feat: add moderation and background fields to ImageRequest struct in dalle.go #1052

feat: add option to allow worker HTTP image requests

README.en.md

@@ -1,10 +1,13 @@
+<p align="right">
+   <a href="./README.md">中文</a> | <strong>English</strong>
+</p>
 <div align="center">
 
 ![new-api](/web/public/logo.png)
 
 # New API
 
-🍥 Next Generation LLM Gateway and AI Asset Management System
+🍥 Next-Generation Large Model Gateway and AI Asset Management System
 
 <a href="https://trendshift.io/repositories/8227" target="_blank"><img src="https://trendshift.io/api/badge/repositories/8227" alt="Calcium-Ion%2Fnew-api | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
 
@@ -33,171 +36,155 @@
 > This is an open-source project developed based on [One API](https://github.com/songquanpeng/one-api)
 
 > [!IMPORTANT]  
-> - Users must comply with OpenAI's [Terms of Use](https://openai.com/policies/terms-of-use) and relevant laws and regulations. Not to be used for illegal purposes.
-> - This project is for personal learning only. Stability is not guaranteed, and no technical support is provided.
+> - This project is for personal learning purposes only, with no guarantee of stability or technical support.
+> - Users must comply with OpenAI's [Terms of Use](https://openai.com/policies/terms-of-use) and **applicable laws and regulations**, and must not use it for illegal purposes.
+> - According to the [《Interim Measures for the Management of Generative Artificial Intelligence Services》](http://www.cac.gov.cn/2023-07/13/c_1690898327029107.htm), please do not provide any unregistered generative AI services to the public in China.
+
+## 📚 Documentation
+
+For detailed documentation, please visit our official Wiki: [https://docs.newapi.pro/](https://docs.newapi.pro/)
 
 ## ✨ Key Features
 
-1. 🎨 New UI interface (some interfaces pending update)
-2. 🌍 Multi-language support (work in progress)
-3. 🎨 Added [Midjourney-Proxy(Plus)](https://github.com/novicezk/midjourney-proxy) interface support, [Integration Guide](Midjourney.md)
-4. 💰 Online recharge support, configurable in system settings:
-    - [x] EasyPay
-5. 🔍 Query usage quota by key:
-    - Works with [neko-api-key-tool](https://github.com/Calcium-Ion/neko-api-key-tool)
-6. 📑 Configurable items per page in pagination
-7. 🔄 Compatible with original One API database (one-api.db)
-8. 💵 Support per-request model pricing, configurable in System Settings - Operation Settings
-9. ⚖️ Support channel **weighted random** selection
-10. 📈 Data dashboard (console)
-11. 🔒 Configurable model access per token
-12. 🤖 Telegram authorization login support:
-    1. System Settings - Configure Login Registration - Allow Telegram Login
-    2. Send /setdomain command to [@Botfather](https://t.me/botfather)
-    3. Select your bot, then enter http(s)://your-website/login
-    4. Telegram Bot name is the bot username without @
-13. 🎵 Added [Suno API](https://github.com/Suno-API/Suno-API) interface support, [Integration Guide](Suno.md)
-14. 🔄 Support for Rerank models, compatible with Cohere and Jina, can integrate with Dify, [Integration Guide](Rerank.md)
-15. ⚡ **[OpenAI Realtime API](https://platform.openai.com/docs/guides/realtime/integration)** - Support for OpenAI's Realtime API, including Azure channels
-16. 🧠 Support for setting reasoning effort through model name suffix:
-    - Add suffix `-high` to set high reasoning effort (e.g., `o3-mini-high`)
-    - Add suffix `-medium` to set medium reasoning effort
-    - Add suffix `-low` to set low reasoning effort
-17. 🔄 Thinking to content option `thinking_to_content` in `Channel->Edit->Channel Extra Settings`, default is `false`, when `true`, the `reasoning_content` of the thinking content will be converted to `<think>` tags and concatenated to the content returned.
-18. 🔄 Model rate limit, support setting total request limit and successful request limit in `System Settings->Rate Limit Settings`
-19. 💰 Cache billing support, when enabled can charge a configurable ratio for cache hits:
-    1. Set `Prompt Cache Ratio` in `System Settings -> Operation Settings`
-    2. Set `Prompt Cache Ratio` in channel settings, range 0-1 (e.g., 0.5 means 50% charge on cache hits)
+New API offers a wide range of features, please refer to [Features Introduction](https://docs.newapi.pro/wiki/features-introduction) for details:
+
+1. 🎨 Brand new UI interface
+2. 🌍 Multi-language support
+3. 💰 Online recharge functionality (YiPay)
+4. 🔍 Support for querying usage quotas with keys (works with [neko-api-key-tool](https://github.com/Calcium-Ion/neko-api-key-tool))
+5. 🔄 Compatible with the original One API database
+6. 💵 Support for pay-per-use model pricing
+7. ⚖️ Support for weighted random channel selection
+8. 📈 Data dashboard (console)
+9. 🔒 Token grouping and model restrictions
+10. 🤖 Support for more authorization login methods (LinuxDO, Telegram, OIDC)
+11. 🔄 Support for Rerank models (Cohere and Jina), [API Documentation](https://docs.newapi.pro/api/jinaai-rerank)
+12. ⚡ Support for OpenAI Realtime API (including Azure channels), [API Documentation](https://docs.newapi.pro/api/openai-realtime)
+13. ⚡ Support for Claude Messages format, [API Documentation](https://docs.newapi.pro/api/anthropic-chat)
+14. Support for entering chat interface via /chat2link route
+15. 🧠 Support for setting reasoning effort through model name suffixes:
+    1. OpenAI o-series models
+        - Add `-high` suffix for high reasoning effort (e.g.: `o3-mini-high`)
+        - Add `-medium` suffix for medium reasoning effort (e.g.: `o3-mini-medium`)
+        - Add `-low` suffix for low reasoning effort (e.g.: `o3-mini-low`)
+    2. Claude thinking models
+        - Add `-thinking` suffix to enable thinking mode (e.g.: `claude-3-7-sonnet-20250219-thinking`)
+16. 🔄 Thinking-to-content functionality
+17. 🔄 Model rate limiting for users
+18. 💰 Cache billing support, which allows billing at a set ratio when cache is hit:
+    1. Set the `Prompt Cache Ratio` option in `System Settings-Operation Settings`
+    2. Set `Prompt Cache Ratio` in the channel, range 0-1, e.g., setting to 0.5 means billing at 50% when cache is hit
     3. Supported channels:
         - [x] OpenAI
-        - [x] Azure 
+        - [x] Azure
         - [x] DeepSeek
-        - [ ] Claude
+        - [x] Claude
 
 ## Model Support
-This version additionally supports:
-1. Third-party model **gpts** (gpt-4-gizmo-*)
-2. [Midjourney-Proxy(Plus)](https://github.com/novicezk/midjourney-proxy) interface, [Integration Guide](Midjourney.md)
-3. Custom channels with full API URL support
-4. [Suno API](https://github.com/Suno-API/Suno-API) interface, [Integration Guide](Suno.md)
-5. Rerank models, supporting [Cohere](https://cohere.ai/) and [Jina](https://jina.ai/), [Integration Guide](Rerank.md)
-6. Dify
 
-You can add custom models gpt-4-gizmo-* in channels. These are third-party models and cannot be called with official OpenAI keys.
+This version supports multiple models, please refer to [API Documentation-Relay Interface](https://docs.newapi.pro/api) for details:
 
-## Additional Configurations Beyond One API
-- `GENERATE_DEFAULT_TOKEN`: Generate initial token for new users, default `false`
-- `STREAMING_TIMEOUT`: Set streaming response timeout, default 60 seconds
-- `DIFY_DEBUG`: Output workflow and node info to client for Dify channel, default `true`
-- `FORCE_STREAM_OPTION`: Override client stream_options parameter, default `true`
-- `GET_MEDIA_TOKEN`: Calculate image tokens, default `true`
-- `GET_MEDIA_TOKEN_NOT_STREAM`: Calculate image tokens in non-stream mode, default `true`
-- `UPDATE_TASK`: Update async tasks (Midjourney, Suno), default `true`
-- `GEMINI_MODEL_MAP`: Specify Gemini model versions (v1/v1beta), format: "model:version", comma-separated
-- `COHERE_SAFETY_SETTING`: Cohere model [safety settings](https://docs.cohere.com/docs/safety-modes#overview), options: `NONE`, `CONTEXTUAL`, `STRICT`, default `NONE`
-- `GEMINI_VISION_MAX_IMAGE_NUM`: Gemini model maximum image number, default `16`, set to `-1` to disable
-- `MAX_FILE_DOWNLOAD_MB`: Maximum file download size in MB, default `20`
-- `CRYPTO_SECRET`: Encryption key for encrypting database content
-- `AZURE_DEFAULT_API_VERSION`: Azure channel default API version, if not specified in channel settings, use this version, default `2024-12-01-preview`
-- `NOTIFICATION_LIMIT_DURATION_MINUTE`: Duration of notification limit in minutes, default `10`
-- `NOTIFY_LIMIT_COUNT`: Maximum number of user notifications in the specified duration, default `2`
+1. Third-party models **gpts** (gpt-4-gizmo-*)
+2. Third-party channel [Midjourney-Proxy(Plus)](https://github.com/novicezk/midjourney-proxy) interface, [API Documentation](https://docs.newapi.pro/api/midjourney-proxy-image)
+3. Third-party channel [Suno API](https://github.com/Suno-API/Suno-API) interface, [API Documentation](https://docs.newapi.pro/api/suno-music)
+4. Custom channels, supporting full call address input
+5. Rerank models ([Cohere](https://cohere.ai/) and [Jina](https://jina.ai/)), [API Documentation](https://docs.newapi.pro/api/jinaai-rerank)
+6. Claude Messages format, [API Documentation](https://docs.newapi.pro/api/anthropic-chat)
+7. Dify, currently only supports chatflow
+
+## Environment Variable Configuration
+
+For detailed configuration instructions, please refer to [Installation Guide-Environment Variables Configuration](https://docs.newapi.pro/installation/environment-variables):
+
+- `GENERATE_DEFAULT_TOKEN`: Whether to generate initial tokens for newly registered users, default is `false`
+- `STREAMING_TIMEOUT`: Streaming response timeout, default is 60 seconds
+- `DIFY_DEBUG`: Whether to output workflow and node information for Dify channels, default is `true`
+- `FORCE_STREAM_OPTION`: Whether to override client stream_options parameter, default is `true`
+- `GET_MEDIA_TOKEN`: Whether to count image tokens, default is `true`
+- `GET_MEDIA_TOKEN_NOT_STREAM`: Whether to count image tokens in non-streaming cases, default is `true`
+- `UPDATE_TASK`: Whether to update asynchronous tasks (Midjourney, Suno), default is `true`
+- `COHERE_SAFETY_SETTING`: Cohere model safety settings, options are `NONE`, `CONTEXTUAL`, `STRICT`, default is `NONE`
+- `GEMINI_VISION_MAX_IMAGE_NUM`: Maximum number of images for Gemini models, default is `16`
+- `MAX_FILE_DOWNLOAD_MB`: Maximum file download size in MB, default is `20`
+- `CRYPTO_SECRET`: Encryption key used for encrypting database content
+- `AZURE_DEFAULT_API_VERSION`: Azure channel default API version, default is `2025-04-01-preview`
+- `NOTIFICATION_LIMIT_DURATION_MINUTE`: Notification limit duration, default is `10` minutes
+- `NOTIFY_LIMIT_COUNT`: Maximum number of user notifications within the specified duration, default is `2`
 
 ## Deployment
 
+For detailed deployment guides, please refer to [Installation Guide-Deployment Methods](https://docs.newapi.pro/installation):
+
 > [!TIP]
-> Latest Docker image: `calciumion/new-api:latest`  
-> Default account: root, password: 123456
+> Latest Docker image: `calciumion/new-api:latest`
 
-### Multi-Server Deployment
-- Must set `SESSION_SECRET` environment variable, otherwise login state will not be consistent across multiple servers.
-- If using a public Redis, must set `CRYPTO_SECRET` environment variable, otherwise Redis content will not be able to be obtained in multi-server deployment.
+### Multi-machine Deployment Considerations
+- Environment variable `SESSION_SECRET` must be set, otherwise login status will be inconsistent across multiple machines
+- If sharing Redis, `CRYPTO_SECRET` must be set, otherwise Redis content cannot be accessed across multiple machines
 
-### Requirements
-- Local database (default): SQLite (Docker deployment must mount `/data` directory)
-- Remote database: MySQL >= 5.7.8, PgSQL >= 9.6
+### Deployment Requirements
+- Local database (default): SQLite (Docker deployment must mount the `/data` directory)
+- Remote database: MySQL version >= 5.7.8, PgSQL version >= 9.6
 
-### Deployment with BT Panel
-Install BT Panel (**version 9.2.0** or above) from [BT Panel Official Website](https://www.bt.cn/new/download.html), choose the stable version script to download and install.  
-After installation, log in to BT Panel and click Docker in the menu bar. First-time access will prompt to install Docker service. Click Install Now and follow the prompts to complete installation.  
-After installation, find **New-API** in the app store, click install, configure basic options to complete installation.  
-[Pictorial Guide](BT.md)
+### Deployment Methods
 
-### Docker Deployment
+#### Using BaoTa Panel Docker Feature
+Install BaoTa Panel (version **9.2.0** or above), find **New-API** in the application store and install it.
+[Tutorial with images](./docs/BT.md)
 
-### Using Docker Compose (Recommended)
+#### Using Docker Compose (Recommended)
 ```shell
-# Clone project
+# Download the project
 git clone https://github.com/Calcium-Ion/new-api.git
 cd new-api
 # Edit docker-compose.yml as needed
-# nano docker-compose.yml
-# vim docker-compose.yml
 # Start
 docker-compose up -d
 ```
 
-#### Update Version
+#### Using Docker Image Directly
 ```shell
-docker-compose pull
-docker-compose up -d
-```
-
-### Direct Docker Image Usage
-```shell
-# SQLite deployment:
+# Using SQLite
 docker run --name new-api -d --restart always -p 3000:3000 -e TZ=Asia/Shanghai -v /home/ubuntu/data/new-api:/data calciumion/new-api:latest
 
-# MySQL deployment (add -e SQL_DSN="root:123456@tcp(localhost:3306)/oneapi"), modify database connection parameters as needed
-# Example:
+# Using MySQL
 docker run --name new-api -d --restart always -p 3000:3000 -e SQL_DSN="root:123456@tcp(localhost:3306)/oneapi" -e TZ=Asia/Shanghai -v /home/ubuntu/data/new-api:/data calciumion/new-api:latest
 ```
 
-#### Update Version
-```shell
-# Pull the latest image
-docker pull calciumion/new-api:latest
-# Stop and remove the old container
-docker stop new-api
-docker rm new-api
-# Run the new container with the same parameters as before
-docker run --name new-api -d --restart always -p 3000:3000 -e TZ=Asia/Shanghai -v /home/ubuntu/data/new-api:/data calciumion/new-api:latest
-```
+## Channel Retry and Cache
+Channel retry functionality has been implemented, you can set the number of retries in `Settings->Operation Settings->General Settings`. It is **recommended to enable caching**.
 
-Alternatively, you can use Watchtower for automatic updates (not recommended, may cause database incompatibility):
-```shell
-docker run --rm -v /var/run/docker.sock:/var/run/docker.sock containrrr/watchtower -cR
-```
+### Cache Configuration Method
+1. `REDIS_CONN_STRING`: Set Redis as cache
+2. `MEMORY_CACHE_ENABLED`: Enable memory cache (no need to set manually if Redis is set)
 
-## Channel Retry
-Channel retry is implemented, configurable in `Settings->Operation Settings->General Settings`. **Cache recommended**.  
-If retry is enabled, the system will automatically use the next priority channel for the same request after a failed request.
+## API Documentation
 
-### Cache Configuration
-1. `REDIS_CONN_STRING`: Use Redis as cache
-    + Example: `REDIS_CONN_STRING=redis://default:redispw@localhost:49153`
-2. `MEMORY_CACHE_ENABLED`: Enable memory cache, default `false`
-    + Example: `MEMORY_CACHE_ENABLED=true`
+For detailed API documentation, please refer to [API Documentation](https://docs.newapi.pro/api):
 
-### Why Some Errors Don't Retry
-Error codes 400, 504, 524 won't retry
-### To Enable Retry for 400
-In `Channel->Edit`, set `Status Code Override` to:
-```json
-{
-  "400": "500"
-}
-```
-
-## Integration Guides
-- [Midjourney Integration](Midjourney.md)
-- [Suno Integration](Suno.md)
+- [Chat API](https://docs.newapi.pro/api/openai-chat)
+- [Image API](https://docs.newapi.pro/api/openai-image)
+- [Rerank API](https://docs.newapi.pro/api/jinaai-rerank)
+- [Realtime API](https://docs.newapi.pro/api/openai-realtime)
+- [Claude Chat API (messages)](https://docs.newapi.pro/api/anthropic-chat)
 
 ## Related Projects
 - [One API](https://github.com/songquanpeng/one-api): Original project
 - [Midjourney-Proxy](https://github.com/novicezk/midjourney-proxy): Midjourney interface support
-- [chatnio](https://github.com/Deeptrain-Community/chatnio): Next-gen AI B/C solution
-- [neko-api-key-tool](https://github.com/Calcium-Ion/neko-api-key-tool): Query usage quota by key
+- [chatnio](https://github.com/Deeptrain-Community/chatnio): Next-generation AI one-stop B/C-end solution
+- [neko-api-key-tool](https://github.com/Calcium-Ion/neko-api-key-tool): Query usage quota with key
+
+Other projects based on New API:
+- [new-api-horizon](https://github.com/Calcium-Ion/new-api-horizon): High-performance optimized version of New API
+- [VoAPI](https://github.com/VoAPI/VoAPI): Frontend beautified version based on New API
+
+## Help and Support
+
+If you have any questions, please refer to [Help and Support](https://docs.newapi.pro/support):
+- [Community Interaction](https://docs.newapi.pro/support/community-interaction)
+- [Issue Feedback](https://docs.newapi.pro/support/feedback-issues)
+- [FAQ](https://docs.newapi.pro/support/faq)
 
 ## 🌟 Star History
 
-[![Star History Chart](https://api.star-history.com/svg?repos=Calcium-Ion/new-api&type=Date)](https://star-history.com/#Calcium-Ion/new-api&Date)
+[![Star History Chart](https://api.star-history.com/svg?repos=Calcium-Ion/new-api&type=Date)](https://star-history.com/#Calcium-Ion/new-api&Date)

README.md

@@ -107,7 +107,7 @@ New API提供了丰富的功能，详细特性请参考[特性说明](https://do
 - `GEMINI_VISION_MAX_IMAGE_NUM`：Gemini模型最大图片数量，默认 `16`
 - `MAX_FILE_DOWNLOAD_MB`: 最大文件下载大小，单位MB，默认 `20`
 - `CRYPTO_SECRET`：加密密钥，用于加密数据库内容
-- `AZURE_DEFAULT_API_VERSION`：Azure渠道默认API版本，默认 `2024-12-01-preview`
+- `AZURE_DEFAULT_API_VERSION`：Azure渠道默认API版本，默认 `2025-04-01-preview`
 - `NOTIFICATION_LIMIT_DURATION_MINUTE`：通知限制持续时间，默认 `10`分钟
 - `NOTIFY_LIMIT_COUNT`：用户通知在指定持续时间内的最大数量，默认 `2`
 
@@ -130,7 +130,7 @@ New API提供了丰富的功能，详细特性请参考[特性说明](https://do
 
 #### 使用宝塔面板Docker功能部署
 安装宝塔面板（**9.2.0版本**及以上），在应用商店中找到**New-API**安装即可。
-[图文教程](BT.md)
+[图文教程](./docs/BT.md)
 
 #### 使用Docker Compose部署（推荐）
 ```shell

common/constants.go

@@ -62,6 +62,10 @@ var EmailDomainWhitelist = []string{
 	"yahoo.com",
 	"foxmail.com",
 }
+var EmailLoginAuthServerList = []string{
+	"smtp.sendcloud.net",
+	"smtp.azurecomm.net",
+}
 
 var DebugEnabled bool
 var MemoryCacheEnabled bool

common/email.go

@@ -5,6 +5,7 @@ import (
 	"encoding/base64"
 	"fmt"
 	"net/smtp"
+	"slices"
 	"strings"
 	"time"
 )
@@ -79,7 +80,7 @@ func SendEmail(subject string, receiver string, content string) error {
 		if err != nil {
 			return err
 		}
-	} else if isOutlookServer(SMTPAccount) || SMTPServer == "smtp.azurecomm.net" {
+	} else if isOutlookServer(SMTPAccount) || slices.Contains(EmailLoginAuthServerList, SMTPServer) {
 		auth = LoginAuth(SMTPAccount, SMTPToken)
 		err = smtp.SendMail(addr, auth, SMTPFrom, to, mail)
 	} else {

common/limiter/limiter.go

@@ -0,0 +1,89 @@
+package limiter
+
+import (
+	"context"
+	_ "embed"
+	"fmt"
+	"github.com/go-redis/redis/v8"
+	"one-api/common"
+	"sync"
+)
+
+//go:embed lua/rate_limit.lua
+var rateLimitScript string
+
+type RedisLimiter struct {
+	client         *redis.Client
+	limitScriptSHA string
+}
+
+var (
+	instance *RedisLimiter
+	once     sync.Once
+)
+
+func New(ctx context.Context, r *redis.Client) *RedisLimiter {
+	once.Do(func() {
+		// 预加载脚本
+		limitSHA, err := r.ScriptLoad(ctx, rateLimitScript).Result()
+		if err != nil {
+			common.SysLog(fmt.Sprintf("Failed to load rate limit script: %v", err))
+		}
+		instance = &RedisLimiter{
+			client:         r,
+			limitScriptSHA: limitSHA,
+		}
+	})
+
+	return instance
+}
+
+func (rl *RedisLimiter) Allow(ctx context.Context, key string, opts ...Option) (bool, error) {
+	// 默认配置
+	config := &Config{
+		Capacity:  10,
+		Rate:      1,
+		Requested: 1,
+	}
+
+	// 应用选项模式
+	for _, opt := range opts {
+		opt(config)
+	}
+
+	// 执行限流
+	result, err := rl.client.EvalSha(
+		ctx,
+		rl.limitScriptSHA,
+		[]string{key},
+		config.Requested,
+		config.Rate,
+		config.Capacity,
+	).Int()
+
+	if err != nil {
+		return false, fmt.Errorf("rate limit failed: %w", err)
+	}
+	return result == 1, nil
+}
+
+// Config 配置选项模式
+type Config struct {
+	Capacity  int64
+	Rate      int64
+	Requested int64
+}
+
+type Option func(*Config)
+
+func WithCapacity(c int64) Option {
+	return func(cfg *Config) { cfg.Capacity = c }
+}
+
+func WithRate(r int64) Option {
+	return func(cfg *Config) { cfg.Rate = r }
+}
+
+func WithRequested(n int64) Option {
+	return func(cfg *Config) { cfg.Requested = n }
+}

common/limiter/lua/rate_limit.lua

@@ -0,0 +1,44 @@
+-- 令牌桶限流器
+-- KEYS[1]: 限流器唯一标识
+-- ARGV[1]: 请求令牌数 (通常为1)
+-- ARGV[2]: 令牌生成速率 (每秒)
+-- ARGV[3]: 桶容量
+
+local key = KEYS[1]
+local requested = tonumber(ARGV[1])
+local rate = tonumber(ARGV[2])
+local capacity = tonumber(ARGV[3])
+
+-- 获取当前时间（Redis服务器时间）
+local now = redis.call('TIME')
+local nowInSeconds = tonumber(now[1])
+
+-- 获取桶状态
+local bucket = redis.call('HMGET', key, 'tokens', 'last_time')
+local tokens = tonumber(bucket[1])
+local last_time = tonumber(bucket[2])
+
+-- 初始化桶（首次请求或过期）
+if not tokens or not last_time then
+    tokens = capacity
+    last_time = nowInSeconds
+else
+    -- 计算新增令牌
+    local elapsed = nowInSeconds - last_time
+    local add_tokens = elapsed * rate
+    tokens = math.min(capacity, tokens + add_tokens)
+    last_time = nowInSeconds
+end
+
+-- 判断是否允许请求
+local allowed = false
+if tokens >= requested then
+    tokens = tokens - requested
+    allowed = true
+end
+
+---- 更新桶状态并设置过期时间
+redis.call('HMSET', key, 'tokens', tokens, 'last_time', last_time)
+--redis.call('EXPIRE', key, math.ceil(capacity / rate) + 60) -- 适当延长过期时间
+
+return allowed and 1 or 0

common/utils.go

@@ -7,7 +7,6 @@ import (
 	"encoding/base64"
 	"encoding/json"
 	"fmt"
-	"github.com/pkg/errors"
 	"html/template"
 	"io"
 	"log"
@@ -22,6 +21,7 @@ import (
 	"time"
 
 	"github.com/google/uuid"
+	"github.com/pkg/errors"
 )
 
 func OpenBrowser(url string) {

constant/azure.go

@@ -0,0 +1,5 @@
+package constant
+
+import "time"
+
+var AzureNoRemoveDotTime = time.Date(2025, time.May, 10, 0, 0, 0, 0, time.UTC).Unix()

constant/env.go

@@ -16,6 +16,7 @@ var GeminiVisionMaxImageNum int
 var NotifyLimitCount int
 var NotificationLimitDurationMinute int
 var GenerateDefaultToken bool
+var ErrorLogEnabled bool
 
 //var GeminiModelMap = map[string]string{
 //	"gemini-1.0-pro": "v1",
@@ -30,12 +31,14 @@ func InitEnv() {
 	GetMediaToken = common.GetEnvOrDefaultBool("GET_MEDIA_TOKEN", true)
 	GetMediaTokenNotStream = common.GetEnvOrDefaultBool("GET_MEDIA_TOKEN_NOT_STREAM", true)
 	UpdateTask = common.GetEnvOrDefaultBool("UPDATE_TASK", true)
-	AzureDefaultAPIVersion = common.GetEnvOrDefaultString("AZURE_DEFAULT_API_VERSION", "2024-12-01-preview")
+	AzureDefaultAPIVersion = common.GetEnvOrDefaultString("AZURE_DEFAULT_API_VERSION", "2025-04-01-preview")
 	GeminiVisionMaxImageNum = common.GetEnvOrDefault("GEMINI_VISION_MAX_IMAGE_NUM", 16)
 	NotifyLimitCount = common.GetEnvOrDefault("NOTIFY_LIMIT_COUNT", 2)
 	NotificationLimitDurationMinute = common.GetEnvOrDefault("NOTIFICATION_LIMIT_DURATION_MINUTE", 10)
 	// GenerateDefaultToken 是否生成初始令牌，默认关闭。
 	GenerateDefaultToken = common.GetEnvOrDefaultBool("GENERATE_DEFAULT_TOKEN", false)
+	// 是否启用错误日志
+	ErrorLogEnabled = common.GetEnvOrDefaultBool("ERROR_LOG_ENABLED", false)
 
 	//modelVersionMapStr := strings.TrimSpace(os.Getenv("GEMINI_MODEL_MAP"))
 	//if modelVersionMapStr == "" {

controller/channel-billing.go

@@ -108,6 +108,13 @@ type DeepSeekUsageResponse struct {
 	} `json:"balance_infos"`
 }
 
+type OpenRouterCreditResponse struct {
+	Data struct {
+		TotalCredits float64 `json:"total_credits"`
+		TotalUsage   float64 `json:"total_usage"`
+	} `json:"data"`
+}
+
 // GetAuthHeader get auth header
 func GetAuthHeader(token string) http.Header {
 	h := http.Header{}
@@ -281,6 +288,22 @@ func updateChannelAIGC2DBalance(channel *model.Channel) (float64, error) {
 	return response.TotalAvailable, nil
 }
 
+func updateChannelOpenRouterBalance(channel *model.Channel) (float64, error) {
+	url := "https://openrouter.ai/api/v1/credits"
+	body, err := GetResponseBody("GET", url, channel, GetAuthHeader(channel.Key))
+	if err != nil {
+		return 0, err
+	}
+	response := OpenRouterCreditResponse{}
+	err = json.Unmarshal(body, &response)
+	if err != nil {
+		return 0, err
+	}
+	balance := response.Data.TotalCredits - response.Data.TotalUsage
+	channel.UpdateBalance(balance)
+	return balance, nil
+}
+
 func updateChannelBalance(channel *model.Channel) (float64, error) {
 	baseURL := common.ChannelBaseURLs[channel.Type]
 	if channel.GetBaseURL() == "" {
@@ -307,6 +330,8 @@ func updateChannelBalance(channel *model.Channel) (float64, error) {
 		return updateChannelSiliconFlowBalance(channel)
 	case common.ChannelTypeDeepSeek:
 		return updateChannelDeepSeekBalance(channel)
+	case common.ChannelTypeOpenRouter:
+		return updateChannelOpenRouterBalance(channel)
 	default:
 		return 0, errors.New("尚未实现")
 	}

controller/channel-test.go

@@ -103,7 +103,10 @@ func testChannel(channel *model.Channel, testModel string) (err error, openAIErr
 	}
 
 	request := buildTestRequest(testModel)
-	common.SysLog(fmt.Sprintf("testing channel %d with model %s , info %v ", channel.Id, testModel, info))
+	// 创建一个用于日志的 info 副本，移除 ApiKey
+	logInfo := *info
+	logInfo.ApiKey = ""
+	common.SysLog(fmt.Sprintf("testing channel %d with model %s , info %+v ", channel.Id, testModel, logInfo))
 
 	priceData, err := helper.ModelPriceHelper(c, info, 0, int(request.MaxTokens))
 	if err != nil {

controller/log.go

@@ -196,7 +196,7 @@ func DeleteHistoryLogs(c *gin.Context) {
 		})
 		return
 	}
-	count, err := model.DeleteOldLog(targetTimestamp)
+	count, err := model.DeleteOldLog(c.Request.Context(), targetTimestamp, 100)
 	if err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,

controller/option.go

@@ -110,6 +110,15 @@ func UpdateOption(c *gin.Context) {
 			})
 			return
 		}
+	case "ModelRequestRateLimitGroup":
+		err = setting.CheckModelRequestRateLimitGroup(option.Value)
+		if err != nil {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": err.Error(),
+			})
+			return
+		}
 
 	}
 	err = model.UpdateOption(option.Key, option.Value)

controller/relay.go

@@ -4,12 +4,11 @@ import (
 	"bytes"
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
-	"github.com/gorilla/websocket"
 	"io"
 	"log"
 	"net/http"
 	"one-api/common"
+	constant2 "one-api/constant"
 	"one-api/dto"
 	"one-api/middleware"
 	"one-api/model"
@@ -19,12 +18,15 @@ import (
 	"one-api/relay/helper"
 	"one-api/service"
 	"strings"
+
+	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 )
 
 func relayHandler(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode {
 	var err *dto.OpenAIErrorWithStatusCode
 	switch relayMode {
-	case relayconstant.RelayModeImagesGenerations:
+	case relayconstant.RelayModeImagesGenerations, relayconstant.RelayModeImagesEdits:
 		err = relay.ImageHelper(c)
 	case relayconstant.RelayModeAudioSpeech:
 		fallthrough
@@ -36,9 +38,31 @@ func relayHandler(c *gin.Context, relayMode int) *dto.OpenAIErrorWithStatusCode
 		err = relay.RerankHelper(c, relayMode)
 	case relayconstant.RelayModeEmbeddings:
 		err = relay.EmbeddingHelper(c)
+	case relayconstant.RelayModeResponses:
+		err = relay.ResponsesHelper(c)
 	default:
 		err = relay.TextHelper(c)
 	}
+
+	if constant2.ErrorLogEnabled && err != nil {
+		// 保存错误日志到mysql中
+		userId := c.GetInt("id")
+		tokenName := c.GetString("token_name")
+		modelName := c.GetString("original_model")
+		tokenId := c.GetInt("token_id")
+		userGroup := c.GetString("group")
+		channelId := c.GetInt("channel_id")
+		other := make(map[string]interface{})
+		other["error_type"] = err.Error.Type
+		other["error_code"] = err.Error.Code
+		other["status_code"] = err.StatusCode
+		other["channel_id"] = channelId
+		other["channel_name"] = c.GetString("channel_name")
+		other["channel_type"] = c.GetInt("channel_type")
+
+		model.RecordErrorLog(c, userId, channelId, modelName, tokenName, err.Error.Message, tokenId, 0, false, userGroup, other)
+	}
+
 	return err
 }
 

controller/user.go

@@ -592,7 +592,14 @@ func UpdateSelf(c *gin.Context) {
 		user.Password = "" // rollback to what it should be
 		cleanUser.Password = ""
 	}
-	updatePassword := user.Password != ""
+	updatePassword, err := checkUpdatePassword(user.OriginalPassword, user.Password, cleanUser.Id)
+	if err != nil {
+		c.JSON(http.StatusOK, gin.H{
+			"success": false,
+			"message": err.Error(),
+		})
+		return
+	}
 	if err := cleanUser.Update(updatePassword); err != nil {
 		c.JSON(http.StatusOK, gin.H{
 			"success": false,
@@ -608,6 +615,23 @@ func UpdateSelf(c *gin.Context) {
 	return
 }
 
+func checkUpdatePassword(originalPassword string, newPassword string, userId int) (updatePassword bool, err error) {
+	var currentUser *model.User
+	currentUser, err = model.GetUserById(userId, true)
+	if err != nil {
+		return
+	}
+	if !common.ValidatePasswordAndHash(originalPassword, currentUser.Password) {
+		err = fmt.Errorf("原密码错误")
+		return
+	}
+	if newPassword == "" {
+		return
+	}
+	updatePassword = true
+	return
+}
+
 func DeleteUser(c *gin.Context) {
 	id, err := strconv.Atoi(c.Param("id"))
 	if err != nil {

docker-compose.yml

@@ -15,6 +15,7 @@ services:
       - SQL_DSN=root:123456@tcp(mysql:3306)/new-api  # Point to the mysql service
       - REDIS_CONN_STRING=redis://redis
       - TZ=Asia/Shanghai
+      - ERROR_LOG_ENABLED=true # 是否启用错误日志记录
     #      - TIKTOKEN_CACHE_DIR=./tiktoken_cache  # 如果需要使用tiktoken_cache，请取消注释
     #      - SESSION_SECRET=random_string  # 多机部署时设置，必须修改这个随机字符串！！！！！！！
     #      - NODE_TYPE=slave  # Uncomment for slave node in multi-node deployment

docs/installation/BT.md

@@ -1,3 +1,3 @@
-密钥为环境变量SESSION_SECRET
-
-![8285bba413e770fe9620f1bf9b40d44e](https://github.com/user-attachments/assets/7a6fc03e-c457-45e4-b8f9-184508fc26b0)
+密钥为环境变量SESSION_SECRET
+
+![8285bba413e770fe9620f1bf9b40d44e](https://github.com/user-attachments/assets/7a6fc03e-c457-45e4-b8f9-184508fc26b0)

dto/claude.go

@@ -70,8 +70,9 @@ func (c *ClaudeMediaMessage) ParseMediaContent() []ClaudeMediaMessage {
 
 type ClaudeMessageSource struct {
 	Type      string `json:"type"`
-	MediaType string `json:"media_type"`
-	Data      any    `json:"data"`
+	MediaType string `json:"media_type,omitempty"`
+	Data      any    `json:"data,omitempty"`
+	Url       string `json:"url,omitempty"`
 }
 
 type ClaudeMessage struct {

dto/dalle.go

@@ -1,17 +1,16 @@
 package dto
 
-import "encoding/json"
-
 type ImageRequest struct {
-	Model          string          `json:"model"`
-	Prompt         string          `json:"prompt" binding:"required"`
-	N              int             `json:"n,omitempty"`
-	Size           string          `json:"size,omitempty"`
-	Quality        string          `json:"quality,omitempty"`
-	ResponseFormat string          `json:"response_format,omitempty"`
-	Style          string          `json:"style,omitempty"`
-	User           string          `json:"user,omitempty"`
-	ExtraFields    json.RawMessage `json:"extra_fields,omitempty"`
+	Model          string `json:"model"`
+	Prompt         string `json:"prompt" binding:"required"`
+	N              int    `json:"n,omitempty"`
+	Size           string `json:"size,omitempty"`
+	Quality        string `json:"quality,omitempty"`
+	ResponseFormat string `json:"response_format,omitempty"`
+	Style          string `json:"style,omitempty"`
+	User           string `json:"user,omitempty"`
+	Moderation     string `json:"moderation,omitempty"`
+	Background     string `json:"background,omitempty"`
 }
 
 type ImageResponse struct {

dto/openai_request.go

@@ -51,6 +51,7 @@ type GeneralOpenAIRequest struct {
 	Dimensions       int               `json:"dimensions,omitempty"`
 	Modalities       any               `json:"modalities,omitempty"`
 	Audio            any               `json:"audio,omitempty"`
+	EnableThinking   any               `json:"enable_thinking,omitempty"` // ali
 	ExtraBody        any               `json:"extra_body,omitempty"`
 }
 
@@ -113,6 +114,7 @@ type MediaContent struct {
 	ImageUrl   any    `json:"image_url,omitempty"`
 	InputAudio any    `json:"input_audio,omitempty"`
 	File       any    `json:"file,omitempty"`
+	VideoUrl   any    `json:"video_url,omitempty"`
 }
 
 func (m *MediaContent) GetImageMedia() *MessageImageUrl {
@@ -157,11 +159,16 @@ type MessageFile struct {
 	FileId   string `json:"file_id,omitempty"`
 }
 
+type MessageVideoUrl struct {
+	Url string `json:"url"`
+}
+
 const (
 	ContentTypeText       = "text"
 	ContentTypeImageURL   = "image_url"
 	ContentTypeInputAudio = "input_audio"
 	ContentTypeFile       = "file"
+	ContentTypeVideoUrl   = "video_url" // 阿里百炼视频识别
 )
 
 func (m *Message) GetPrefix() bool {
@@ -345,6 +352,15 @@ func (m *Message) ParseContent() []MediaContent {
 						}
 					}
 				}
+			case ContentTypeVideoUrl:
+				if videoUrl, ok := contentItem["video_url"].(string); ok {
+					contentList = append(contentList, MediaContent{
+						Type: ContentTypeVideoUrl,
+						VideoUrl: &MessageVideoUrl{
+							Url: videoUrl,
+						},
+					})
+				}
 			}
 		}
 	}
@@ -354,3 +370,49 @@ func (m *Message) ParseContent() []MediaContent {
 	}
 	return contentList
 }
+
+type OpenAIResponsesRequest struct {
+	Model              string               `json:"model"`
+	Input              json.RawMessage      `json:"input,omitempty"`
+	Include            json.RawMessage      `json:"include,omitempty"`
+	Instructions       json.RawMessage      `json:"instructions,omitempty"`
+	MaxOutputTokens    uint                 `json:"max_output_tokens,omitempty"`
+	Metadata           json.RawMessage      `json:"metadata,omitempty"`
+	ParallelToolCalls  bool                 `json:"parallel_tool_calls,omitempty"`
+	PreviousResponseID string               `json:"previous_response_id,omitempty"`
+	Reasoning          *Reasoning           `json:"reasoning,omitempty"`
+	ServiceTier        string               `json:"service_tier,omitempty"`
+	Store              bool                 `json:"store,omitempty"`
+	Stream             bool                 `json:"stream,omitempty"`
+	Temperature        float64              `json:"temperature,omitempty"`
+	Text               json.RawMessage      `json:"text,omitempty"`
+	ToolChoice         json.RawMessage      `json:"tool_choice,omitempty"`
+	Tools              []ResponsesToolsCall `json:"tools,omitempty"`
+	TopP               float64              `json:"top_p,omitempty"`
+	Truncation         string               `json:"truncation,omitempty"`
+	User               string               `json:"user,omitempty"`
+}
+
+type Reasoning struct {
+	Effort  string `json:"effort,omitempty"`
+	Summary string `json:"summary,omitempty"`
+}
+
+type ResponsesToolsCall struct {
+	Type string `json:"type"`
+	// Web Search
+	UserLocation      json.RawMessage `json:"user_location,omitempty"`
+	SearchContextSize string          `json:"search_context_size,omitempty"`
+	// File Search
+	VectorStoreIds []string        `json:"vector_store_ids,omitempty"`
+	MaxNumResults  uint            `json:"max_num_results,omitempty"`
+	Filters        json.RawMessage `json:"filters,omitempty"`
+	// Computer Use
+	DisplayWidth  uint   `json:"display_width,omitempty"`
+	DisplayHeight uint   `json:"display_height,omitempty"`
+	Environment   string `json:"environment,omitempty"`
+	// Function
+	Name        string          `json:"name,omitempty"`
+	Description string          `json:"description,omitempty"`
+	Parameters  json.RawMessage `json:"parameters,omitempty"`
+}

dto/openai_response.go

@@ -1,5 +1,7 @@
 package dto
 
+import "encoding/json"
+
 type SimpleResponse struct {
 	Usage `json:"usage"`
 	Error *OpenAIError `json:"error"`
@@ -166,12 +168,16 @@ type CompletionsStreamResponse struct {
 }
 
 type Usage struct {
-	PromptTokens           int                `json:"prompt_tokens"`
-	CompletionTokens       int                `json:"completion_tokens"`
-	TotalTokens            int                `json:"total_tokens"`
-	PromptCacheHitTokens   int                `json:"prompt_cache_hit_tokens,omitempty"`
+	PromptTokens         int `json:"prompt_tokens"`
+	CompletionTokens     int `json:"completion_tokens"`
+	TotalTokens          int `json:"total_tokens"`
+	PromptCacheHitTokens int `json:"prompt_cache_hit_tokens,omitempty"`
+
 	PromptTokensDetails    InputTokenDetails  `json:"prompt_tokens_details"`
 	CompletionTokenDetails OutputTokenDetails `json:"completion_tokens_details"`
+	InputTokens            int                `json:"input_tokens"`
+	OutputTokens           int                `json:"output_tokens"`
+	InputTokensDetails     *InputTokenDetails `json:"input_tokens_details"`
 }
 
 type InputTokenDetails struct {
@@ -187,3 +193,68 @@ type OutputTokenDetails struct {
 	AudioTokens     int `json:"audio_tokens"`
 	ReasoningTokens int `json:"reasoning_tokens"`
 }
+
+type OpenAIResponsesResponse struct {
+	ID                 string               `json:"id"`
+	Object             string               `json:"object"`
+	CreatedAt          int                  `json:"created_at"`
+	Status             string               `json:"status"`
+	Error              *OpenAIError         `json:"error,omitempty"`
+	IncompleteDetails  *IncompleteDetails   `json:"incomplete_details,omitempty"`
+	Instructions       string               `json:"instructions"`
+	MaxOutputTokens    int                  `json:"max_output_tokens"`
+	Model              string               `json:"model"`
+	Output             []ResponsesOutput    `json:"output"`
+	ParallelToolCalls  bool                 `json:"parallel_tool_calls"`
+	PreviousResponseID string               `json:"previous_response_id"`
+	Reasoning          *Reasoning           `json:"reasoning"`
+	Store              bool                 `json:"store"`
+	Temperature        float64              `json:"temperature"`
+	ToolChoice         string               `json:"tool_choice"`
+	Tools              []ResponsesToolsCall `json:"tools"`
+	TopP               float64              `json:"top_p"`
+	Truncation         string               `json:"truncation"`
+	Usage              *Usage               `json:"usage"`
+	User               json.RawMessage      `json:"user"`
+	Metadata           json.RawMessage      `json:"metadata"`
+}
+
+type IncompleteDetails struct {
+	Reasoning string `json:"reasoning"`
+}
+
+type ResponsesOutput struct {
+	Type    string                   `json:"type"`
+	ID      string                   `json:"id"`
+	Status  string                   `json:"status"`
+	Role    string                   `json:"role"`
+	Content []ResponsesOutputContent `json:"content"`
+}
+
+type ResponsesOutputContent struct {
+	Type        string        `json:"type"`
+	Text        string        `json:"text"`
+	Annotations []interface{} `json:"annotations"`
+}
+
+const (
+	BuildInToolWebSearchPreview = "web_search_preview"
+	BuildInToolFileSearch       = "file_search"
+)
+
+const (
+	BuildInCallWebSearchCall = "web_search_call"
+)
+
+const (
+	ResponsesOutputTypeItemAdded = "response.output_item.added"
+	ResponsesOutputTypeItemDone  = "response.output_item.done"
+)
+
+// ResponsesStreamResponse 用于处理 /v1/responses 流式响应
+type ResponsesStreamResponse struct {
+	Type     string                   `json:"type"`
+	Response *OpenAIResponsesResponse `json:"response,omitempty"`
+	Delta    string                   `json:"delta,omitempty"`
+	Item     *ResponsesOutput         `json:"item,omitempty"`
+}

main.go

@@ -74,12 +74,14 @@ func main() {
 	}
 
 	// Initialize model settings
-	operation_setting.InitModelSettings()
+	operation_setting.InitRatioSettings()
 	// Initialize constants
 	constant.InitEnv()
 	// Initialize options
 	model.InitOptionMap()
 
+	service.InitTokenEncoders()
+
 	if common.RedisEnabled {
 		// for compatibility with old versions
 		common.MemoryCacheEnabled = true
@@ -133,8 +135,6 @@ func main() {
 		common.SysLog("pprof enabled")
 	}
 
-	service.InitTokenEncoders()
-
 	// Initialize HTTP server
 	server := gin.New()
 	server.Use(gin.CustomRecovery(func(c *gin.Context, err any) {

middleware/distributor.go

@@ -162,7 +162,7 @@ func getModelRequest(c *gin.Context) (*ModelRequest, bool, error) {
 		}
 		c.Set("platform", string(constant.TaskPlatformSuno))
 		c.Set("relay_mode", relayMode)
-	} else if !strings.HasPrefix(c.Request.URL.Path, "/v1/audio/transcriptions") {
+	} else if !strings.HasPrefix(c.Request.URL.Path, "/v1/audio/transcriptions") && !strings.HasPrefix(c.Request.URL.Path, "/v1/images/edits") {
 		err = common.UnmarshalBodyReusable(c, &modelRequest)
 	}
 	if err != nil {
@@ -184,6 +184,8 @@ func getModelRequest(c *gin.Context) (*ModelRequest, bool, error) {
 	}
 	if strings.HasPrefix(c.Request.URL.Path, "/v1/images/generations") {
 		modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, "dall-e")
+	} else if strings.HasPrefix(c.Request.URL.Path, "/v1/images/edits") {
+		modelRequest.Model = common.GetStringIfEmpty(modelRequest.Model, "gpt-image-1")
 	}
 	if strings.HasPrefix(c.Request.URL.Path, "/v1/audio") {
 		relayMode := relayconstant.RelayModeAudioSpeech
@@ -211,6 +213,7 @@ func SetupContextForSelectedChannel(c *gin.Context, channel *model.Channel, mode
 	c.Set("channel_id", channel.Id)
 	c.Set("channel_name", channel.Name)
 	c.Set("channel_type", channel.Type)
+	c.Set("channel_create_time", channel.CreatedTime)
 	c.Set("channel_setting", channel.GetSetting())
 	c.Set("param_override", channel.GetParamOverride())
 	if nil != channel.OpenAIOrganization && "" != *channel.OpenAIOrganization {

middleware/model-rate-limit.go

@@ -5,6 +5,8 @@ import (
 	"fmt"
 	"net/http"
 	"one-api/common"
+	"one-api/common/limiter"
+	"one-api/constant"
 	"one-api/setting"
 	"strconv"
 	"time"
@@ -78,21 +80,9 @@ func redisRateLimitHandler(duration int64, totalMaxCount, successMaxCount int) g
 		ctx := context.Background()
 		rdb := common.RDB
 
-		// 1. 检查总请求数限制（当totalMaxCount为0时会自动跳过）
-		totalKey := fmt.Sprintf("rateLimit:%s:%s", ModelRequestRateLimitCountMark, userId)
-		allowed, err := checkRedisRateLimit(ctx, rdb, totalKey, totalMaxCount, duration)
-		if err != nil {
-			fmt.Println("检查总请求数限制失败:", err.Error())
-			abortWithOpenAiMessage(c, http.StatusInternalServerError, "rate_limit_check_failed")
-			return
-		}
-		if !allowed {
-			abortWithOpenAiMessage(c, http.StatusTooManyRequests, fmt.Sprintf("您已达到总请求数限制：%d分钟内最多请求%d次，包括失败次数，请检查您的请求是否正确", setting.ModelRequestRateLimitDurationMinutes, totalMaxCount))
-		}
-
-		// 2. 检查成功请求数限制
+		// 1. 检查成功请求数限制
 		successKey := fmt.Sprintf("rateLimit:%s:%s", ModelRequestRateLimitSuccessCountMark, userId)
-		allowed, err = checkRedisRateLimit(ctx, rdb, successKey, successMaxCount, duration)
+		allowed, err := checkRedisRateLimit(ctx, rdb, successKey, successMaxCount, duration)
 		if err != nil {
 			fmt.Println("检查成功请求数限制失败:", err.Error())
 			abortWithOpenAiMessage(c, http.StatusInternalServerError, "rate_limit_check_failed")
@@ -103,8 +93,29 @@ func redisRateLimitHandler(duration int64, totalMaxCount, successMaxCount int) g
 			return
 		}
 
-		// 3. 记录总请求（当totalMaxCount为0时会自动跳过）
-		recordRedisRequest(ctx, rdb, totalKey, totalMaxCount)
+		//2.检查总请求数限制并记录总请求（当totalMaxCount为0时会自动跳过，使用令牌桶限流器
+		if totalMaxCount > 0 {
+			totalKey := fmt.Sprintf("rateLimit:%s", userId)
+			// 初始化
+			tb := limiter.New(ctx, rdb)
+			allowed, err = tb.Allow(
+				ctx,
+				totalKey,
+				limiter.WithCapacity(int64(totalMaxCount)*duration),
+				limiter.WithRate(int64(totalMaxCount)),
+				limiter.WithRequested(duration),
+			)
+
+			if err != nil {
+				fmt.Println("检查总请求数限制失败:", err.Error())
+				abortWithOpenAiMessage(c, http.StatusInternalServerError, "rate_limit_check_failed")
+				return
+			}
+
+			if !allowed {
+				abortWithOpenAiMessage(c, http.StatusTooManyRequests, fmt.Sprintf("您已达到总请求数限制：%d分钟内最多请求%d次，包括失败次数，请检查您的请求是否正确", setting.ModelRequestRateLimitDurationMinutes, totalMaxCount))
+			}
+		}
 
 		// 4. 处理请求
 		c.Next()
@@ -165,6 +176,19 @@ func ModelRequestRateLimit() func(c *gin.Context) {
 		totalMaxCount := setting.ModelRequestRateLimitCount
 		successMaxCount := setting.ModelRequestRateLimitSuccessCount
 
+		// 获取分组
+		group := c.GetString("token_group")
+		if group == "" {
+			group = c.GetString(constant.ContextKeyUserGroup)
+		}
+
+		//获取分组的限流配置
+		groupTotalCount, groupSuccessCount, found := setting.GetGroupRateLimit(group)
+		if found {
+			totalMaxCount = groupTotalCount
+			successMaxCount = groupSuccessCount
+		}
+
 		// 根据存储类型选择并执行限流处理器
 		if common.RedisEnabled {
 			redisRateLimitHandler(duration, totalMaxCount, successMaxCount)(c)

model/cache.go

@@ -84,9 +84,11 @@ func CacheGetRandomSatisfiedChannel(group string, model string, retry int) (*Cha
 	if !common.MemoryCacheEnabled {
 		return GetRandomSatisfiedChannel(group, model, retry)
 	}
+	
 	channelSyncLock.RLock()
-	defer channelSyncLock.RUnlock()
 	channels := group2model2channels[group][model]
+	channelSyncLock.RUnlock()
+	
 	if len(channels) == 0 {
 		return nil, errors.New("channel not found")
 	}

model/channel.go

@@ -119,10 +119,15 @@ func SearchChannels(keyword string, group string, model string, idSort bool) ([]
 
 	// 如果是 PostgreSQL，使用双引号
 	if common.UsingPostgreSQL {
-		keyCol = `"key"`
 		modelsCol = `"models"`
 	}
 
+	baseURLCol := "`base_url`"
+	// 如果是 PostgreSQL，使用双引号
+	if common.UsingPostgreSQL {
+		baseURLCol = `"base_url"`
+	}
+
 	order := "priority desc"
 	if idSort {
 		order = "id desc"
@@ -142,11 +147,11 @@ func SearchChannels(keyword string, group string, model string, idSort bool) ([]
 			// sqlite, PostgreSQL
 			groupCondition = `(',' || ` + groupCol + ` || ',') LIKE ?`
 		}
-		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + modelsCol + ` LIKE ? AND ` + groupCondition
-		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+model+"%", "%,"+group+",%")
+		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ? OR " + baseURLCol + " LIKE ?) AND " + modelsCol + ` LIKE ? AND ` + groupCondition
+		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+keyword+"%", "%"+model+"%", "%,"+group+",%")
 	} else {
-		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + modelsCol + " LIKE ?"
-		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+model+"%")
+		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ? OR " + baseURLCol + " LIKE ?) AND " + modelsCol + " LIKE ?"
+		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+keyword+"%", "%"+model+"%")
 	}
 
 	// 执行查询
@@ -450,6 +455,12 @@ func SearchTags(keyword string, group string, model string, idSort bool) ([]*str
 		modelsCol = `"models"`
 	}
 
+	baseURLCol := "`base_url`"
+	// 如果是 PostgreSQL，使用双引号
+	if common.UsingPostgreSQL {
+		baseURLCol = `"base_url"`
+	}
+
 	order := "priority desc"
 	if idSort {
 		order = "id desc"
@@ -469,11 +480,11 @@ func SearchTags(keyword string, group string, model string, idSort bool) ([]*str
 			// sqlite, PostgreSQL
 			groupCondition = `(',' || ` + groupCol + ` || ',') LIKE ?`
 		}
-		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + modelsCol + ` LIKE ? AND ` + groupCondition
-		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+model+"%", "%,"+group+",%")
+		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ? OR " + baseURLCol + " LIKE ?) AND " + modelsCol + ` LIKE ? AND ` + groupCondition
+		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+keyword+"%", "%"+model+"%", "%,"+group+",%")
 	} else {
-		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ?) AND " + modelsCol + " LIKE ?"
-		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+model+"%")
+		whereClause = "(id = ? OR name LIKE ? OR " + keyCol + " = ? OR " + baseURLCol + " LIKE ?) AND " + modelsCol + " LIKE ?"
+		args = append(args, common.String2Int(keyword), "%"+keyword+"%", keyword, "%"+keyword+"%", "%"+model+"%")
 	}
 
 	subQuery := baseQuery.Where(whereClause, args...).

model/log.go

@@ -1,6 +1,7 @@
 package model
 
 import (
+	"context"
 	"fmt"
 	"one-api/common"
 	"os"
@@ -40,6 +41,7 @@ const (
 	LogTypeConsume
 	LogTypeManage
 	LogTypeSystem
+	LogTypeError
 )
 
 func formatUserLogs(logs []*Log) {
@@ -88,6 +90,35 @@ func RecordLog(userId int, logType int, content string) {
 	}
 }
 
+func RecordErrorLog(c *gin.Context, userId int, channelId int, modelName string, tokenName string, content string, tokenId int, useTimeSeconds int,
+	isStream bool, group string, other map[string]interface{}) {
+	common.LogInfo(c, fmt.Sprintf("record error log: userId=%d, channelId=%d, modelName=%s, tokenName=%s, content=%s", userId, channelId, modelName, tokenName, content))
+	username := c.GetString("username")
+	otherStr := common.MapToJsonStr(other)
+	log := &Log{
+		UserId:           userId,
+		Username:         username,
+		CreatedAt:        common.GetTimestamp(),
+		Type:             LogTypeError,
+		Content:          content,
+		PromptTokens:     0,
+		CompletionTokens: 0,
+		TokenName:        tokenName,
+		ModelName:        modelName,
+		Quota:            0,
+		ChannelId:        channelId,
+		TokenId:          tokenId,
+		UseTime:          useTimeSeconds,
+		IsStream:         isStream,
+		Group:            group,
+		Other:            otherStr,
+	}
+	err := LOG_DB.Create(log).Error
+	if err != nil {
+		common.LogError(c, "failed to record log: "+err.Error())
+	}
+}
+
 func RecordConsumeLog(c *gin.Context, userId int, channelId int, promptTokens int, completionTokens int,
 	modelName string, tokenName string, quota int, content string, tokenId int, userQuota int, useTimeSeconds int,
 	isStream bool, group string, other map[string]interface{}) {
@@ -310,7 +341,25 @@ func SumUsedToken(logType int, startTimestamp int64, endTimestamp int64, modelNa
 	return token
 }
 
-func DeleteOldLog(targetTimestamp int64) (int64, error) {
-	result := LOG_DB.Where("created_at < ?", targetTimestamp).Delete(&Log{})
-	return result.RowsAffected, result.Error
+func DeleteOldLog(ctx context.Context, targetTimestamp int64, limit int) (int64, error) {
+	var total int64 = 0
+
+	for {
+		if nil != ctx.Err() {
+			return total, ctx.Err()
+		}
+
+		result := LOG_DB.Where("created_at < ?", targetTimestamp).Limit(limit).Delete(&Log{})
+		if nil != result.Error {
+			return total, result.Error
+		}
+
+		total += result.RowsAffected
+
+		if result.RowsAffected < int64(limit) {
+			break
+		}
+	}
+
+	return total, nil
 }

model/option.go

@@ -67,6 +67,7 @@ func InitOptionMap() {
 	common.OptionMap["ServerAddress"] = ""
 	common.OptionMap["WorkerUrl"] = setting.WorkerUrl
 	common.OptionMap["WorkerValidKey"] = setting.WorkerValidKey
+	common.OptionMap["WorkerAllowHttpImageRequestEnabled"] = strconv.FormatBool(setting.WorkerAllowHttpImageRequestEnabled)
 	common.OptionMap["PayAddress"] = ""
 	common.OptionMap["CustomCallbackAddress"] = ""
 	common.OptionMap["EpayId"] = ""
@@ -92,6 +93,7 @@ func InitOptionMap() {
 	common.OptionMap["ModelRequestRateLimitCount"] = strconv.Itoa(setting.ModelRequestRateLimitCount)
 	common.OptionMap["ModelRequestRateLimitDurationMinutes"] = strconv.Itoa(setting.ModelRequestRateLimitDurationMinutes)
 	common.OptionMap["ModelRequestRateLimitSuccessCount"] = strconv.Itoa(setting.ModelRequestRateLimitSuccessCount)
+	common.OptionMap["ModelRequestRateLimitGroup"] = setting.ModelRequestRateLimitGroup2JSONString()
 	common.OptionMap["ModelRatio"] = operation_setting.ModelRatio2JSONString()
 	common.OptionMap["ModelPrice"] = operation_setting.ModelPrice2JSONString()
 	common.OptionMap["CacheRatio"] = operation_setting.CacheRatio2JSONString()
@@ -256,6 +258,8 @@ func updateOptionMap(key string, value string) (err error) {
 			setting.StopOnSensitiveEnabled = boolValue
 		case "SMTPSSLEnabled":
 			common.SMTPSSLEnabled = boolValue
+		case "WorkerAllowHttpImageRequestEnabled":
+			setting.WorkerAllowHttpImageRequestEnabled = boolValue
 		}
 	}
 	switch key {
@@ -338,6 +342,8 @@ func updateOptionMap(key string, value string) (err error) {
 		setting.ModelRequestRateLimitDurationMinutes, _ = strconv.Atoi(value)
 	case "ModelRequestRateLimitSuccessCount":
 		setting.ModelRequestRateLimitSuccessCount, _ = strconv.Atoi(value)
+	case "ModelRequestRateLimitGroup":
+		err = setting.UpdateModelRequestRateLimitGroupByJSONString(value)
 	case "RetryTimes":
 		common.RetryTimes, _ = strconv.Atoi(value)
 	case "DataExportInterval":

model/user.go

@@ -18,6 +18,7 @@ type User struct {
 	Id               int            `json:"id"`
 	Username         string         `json:"username" gorm:"unique;index" validate:"max=12"`
 	Password         string         `json:"password" gorm:"not null;" validate:"min=8,max=20"`
+	OriginalPassword string         `json:"original_password" gorm:"-:all"` // this field is only for Password change verification, don't save it to database!
 	DisplayName      string         `json:"display_name" gorm:"index" validate:"max=20"`
 	Role             int            `json:"role" gorm:"type:int;default:1"`   // admin, common
 	Status           int            `json:"status" gorm:"type:int;default:1"` // enabled, disabled
@@ -108,7 +109,7 @@ func CheckUserExistOrDeleted(username string, email string) (bool, error) {
 
 func GetMaxUserId() int {
 	var user User
-	DB.Last(&user)
+	DB.Unscoped().Last(&user)
 	return user.Id
 }
 

relay/channel/adapter.go

@@ -1,11 +1,12 @@
 package channel
 
 import (
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	relaycommon "one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor interface {
@@ -18,6 +19,7 @@ type Adaptor interface {
 	ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.EmbeddingRequest) (any, error)
 	ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error)
 	ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error)
+	ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error)
 	DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error)
 	DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode)
 	GetModelList() []string

relay/channel/ali/adaptor.go

@@ -3,7 +3,6 @@ package ali
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,8 @@ import (
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -79,6 +80,11 @@ func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInf
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/ali/constants.go

@@ -1,7 +1,12 @@
 package ali
 
 var ModelList = []string{
-	"qwen-turbo", "qwen-plus", "qwen-max", "qwen-max-longcontext",
+	"qwen-turbo",
+	"qwen-plus",
+	"qwen-max",
+	"qwen-max-longcontext",
+	"qwq-32b",
+	"qwen3-235b-a22b",
 	"text-embedding-v1",
 }
 

relay/channel/api_request.go

@@ -1,16 +1,23 @@
 package channel
 
 import (
+	"context"
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
-	"github.com/gorilla/websocket"
 	"io"
 	"net/http"
 	common2 "one-api/common"
 	"one-api/relay/common"
 	"one-api/relay/constant"
+	"one-api/relay/helper"
 	"one-api/service"
+	"one-api/setting/operation_setting"
+	"sync"
+	"time"
+
+	"github.com/bytedance/gopkg/util/gopool"
+	"github.com/gin-gonic/gin"
+	"github.com/gorilla/websocket"
 )
 
 func SetupApiRequestHeader(info *common.RelayInfo, c *gin.Context, req *http.Header) {
@@ -55,6 +62,9 @@ func DoFormRequest(a Adaptor, c *gin.Context, info *common.RelayInfo, requestBod
 	if err != nil {
 		return nil, fmt.Errorf("get request url failed: %w", err)
 	}
+	if common2.DebugEnabled {
+		println("fullRequestURL:", fullRequestURL)
+	}
 	req, err := http.NewRequest(c.Request.Method, fullRequestURL, requestBody)
 	if err != nil {
 		return nil, fmt.Errorf("new request failed: %w", err)
@@ -105,7 +115,62 @@ func doRequest(c *gin.Context, req *http.Request, info *common.RelayInfo) (*http
 	} else {
 		client = service.GetHttpClient()
 	}
+	// 流式请求 ping 保活
+	var stopPinger func()
+	generalSettings := operation_setting.GetGeneralSetting()
+	pingEnabled := generalSettings.PingIntervalEnabled
+	var pingerWg sync.WaitGroup
+	if info.IsStream {
+		helper.SetEventStreamHeaders(c)
+		pingInterval := time.Duration(generalSettings.PingIntervalSeconds) * time.Second
+		var pingerCtx context.Context
+		pingerCtx, stopPinger = context.WithCancel(c.Request.Context())
+
+		if pingEnabled {
+			pingerWg.Add(1)
+			gopool.Go(func() {
+				defer pingerWg.Done()
+				if pingInterval <= 0 {
+					pingInterval = helper.DefaultPingInterval
+				}
+
+				ticker := time.NewTicker(pingInterval)
+				defer ticker.Stop()
+				var pingMutex sync.Mutex
+				if common2.DebugEnabled {
+					println("SSE ping goroutine started")
+				}
+
+				for {
+					select {
+					case <-ticker.C:
+						pingMutex.Lock()
+						err2 := helper.PingData(c)
+						pingMutex.Unlock()
+						if err2 != nil {
+							common2.LogError(c, "SSE ping error: "+err.Error())
+							return
+						}
+						if common2.DebugEnabled {
+							println("SSE ping data sent.")
+						}
+					case <-pingerCtx.Done():
+						if common2.DebugEnabled {
+							println("SSE ping goroutine stopped.")
+						}
+						return
+					}
+				}
+			})
+		}
+	}
+
 	resp, err := client.Do(req)
+	// request结束后停止ping
+	if info.IsStream && pingEnabled {
+		stopPinger()
+		pingerWg.Wait()
+	}
 	if err != nil {
 		return nil, err
 	}

relay/channel/aws/adaptor.go

@@ -2,13 +2,14 @@ package aws
 
 import (
 	"errors"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel/claude"
 	relaycommon "one-api/relay/common"
 	"one-api/setting/model_setting"
+
+	"github.com/gin-gonic/gin"
 )
 
 const (
@@ -74,6 +75,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return nil, nil
 }

relay/channel/baidu/adaptor.go

@@ -3,7 +3,6 @@ package baidu
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,8 @@ import (
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -130,6 +131,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return baiduEmbeddingRequest, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/baidu_v2/adaptor.go

@@ -3,13 +3,14 @@ package baidu_v2
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -60,6 +61,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/claude/adaptor.go

@@ -3,7 +3,6 @@ package claude
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,8 @@ import (
 	relaycommon "one-api/relay/common"
 	"one-api/setting/model_setting"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 const (
@@ -84,6 +85,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/cloudflare/adaptor.go

@@ -55,6 +55,11 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	}
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/cohere/adaptor.go

@@ -3,13 +3,14 @@ package cohere
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -52,6 +53,11 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	return requestOpenAI2Cohere(*request), nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/deepseek/adaptor.go

@@ -3,7 +3,6 @@ package deepseek
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,9 @@ import (
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
+	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -36,9 +38,13 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
+	fimBaseUrl := info.BaseUrl
+	if !strings.HasSuffix(info.BaseUrl, "/beta") {
+		fimBaseUrl += "/beta"
+	}
 	switch info.RelayMode {
 	case constant.RelayModeCompletions:
-		return fmt.Sprintf("%s/beta/completions", info.BaseUrl), nil
+		return fmt.Sprintf("%s/completions", fimBaseUrl), nil
 	default:
 		return fmt.Sprintf("%s/v1/chat/completions", info.BaseUrl), nil
 	}
@@ -66,6 +72,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/dify/adaptor.go

@@ -3,12 +3,13 @@ package dify
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	relaycommon "one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 const (
@@ -86,6 +87,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/gemini/adaptor.go

@@ -155,6 +155,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return geminiRequest, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/gemini/relay-gemini.go

@@ -391,6 +391,7 @@ func removeAdditionalPropertiesWithDepth(schema interface{}, depth int) interfac
 	}
 	// 删除所有的title字段
 	delete(v, "title")
+	delete(v, "$schema")
 	// 如果type不为object和array，则直接返回
 	if typeVal, exists := v["type"]; !exists || (typeVal != "object" && typeVal != "array") {
 		return schema
@@ -670,6 +671,7 @@ func GeminiChatStreamHandler(c *gin.Context, resp *http.Response, info *relaycom
 			usage.PromptTokens = geminiResponse.UsageMetadata.PromptTokenCount
 			usage.CompletionTokens = geminiResponse.UsageMetadata.CandidatesTokenCount
 			usage.CompletionTokenDetails.ReasoningTokens = geminiResponse.UsageMetadata.ThoughtsTokenCount
+			usage.TotalTokens = geminiResponse.UsageMetadata.TotalTokenCount
 		}
 		err = helper.ObjectData(c, response)
 		if err != nil {
@@ -690,9 +692,8 @@ func GeminiChatStreamHandler(c *gin.Context, resp *http.Response, info *relaycom
 		}
 	}
 
-	usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
 	usage.PromptTokensDetails.TextTokens = usage.PromptTokens
-	//usage.CompletionTokenDetails.TextTokens = usage.CompletionTokens
+	usage.CompletionTokens = usage.TotalTokens - usage.PromptTokens
 
 	if info.ShouldIncludeUsage {
 		response = helper.GenerateFinalUsageResponse(id, createAt, info.UpstreamModelName, *usage)
@@ -740,6 +741,7 @@ func GeminiChatHandler(c *gin.Context, resp *http.Response, info *relaycommon.Re
 	}
 
 	usage.CompletionTokenDetails.ReasoningTokens = geminiResponse.UsageMetadata.ThoughtsTokenCount
+	usage.CompletionTokens = usage.TotalTokens - usage.PromptTokens
 
 	fullTextResponse.Usage = usage
 	jsonResponse, err := json.Marshal(fullTextResponse)

relay/channel/jina/adaptor.go

@@ -3,7 +3,6 @@ package jina
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -12,6 +11,8 @@ import (
 	relaycommon "one-api/relay/common"
 	"one-api/relay/common_handler"
 	"one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -55,6 +56,11 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	return request, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/mistral/adaptor.go

@@ -2,13 +2,14 @@ package mistral
 
 import (
 	"errors"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -59,6 +60,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/mokaai/adaptor.go

@@ -3,7 +3,6 @@ package mokaai
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,8 @@ import (
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -74,6 +75,11 @@ func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dt
 	return nil, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/ollama/adaptor.go

@@ -2,7 +2,6 @@ package ollama
 
 import (
 	"errors"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -10,6 +9,8 @@ import (
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	relayconstant "one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -64,6 +65,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return requestOpenAI2Embeddings(request), nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/openai/adaptor.go

@@ -8,6 +8,7 @@ import (
 	"io"
 	"mime/multipart"
 	"net/http"
+	"net/textproto"
 	"one-api/common"
 	constant2 "one-api/constant"
 	"one-api/dto"
@@ -22,6 +23,7 @@ import (
 	"one-api/relay/common_handler"
 	"one-api/relay/constant"
 	"one-api/service"
+	"path/filepath"
 	"strings"
 
 	"github.com/gin-gonic/gin"
@@ -87,7 +89,10 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 		requestURL = fmt.Sprintf("%s?api-version=%s", requestURL, apiVersion)
 		task := strings.TrimPrefix(requestURL, "/v1/")
 		model_ := info.UpstreamModelName
-		model_ = strings.Replace(model_, ".", "", -1)
+		// 2025年5月10日后创建的渠道不移除.
+		if info.ChannelCreateTime < constant2.AzureNoRemoveDotTime {
+			model_ = strings.Replace(model_, ".", "", -1)
+		}
 		// https://github.com/songquanpeng/one-api/issues/67
 		requestURL = fmt.Sprintf("/openai/deployments/%s/%s", model_, task)
 		if info.RelayMode == constant.RelayModeRealtime {
@@ -167,7 +172,7 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 		info.UpstreamModelName = request.Model
 
 		// o系列模型developer适配（o1-mini除外）
-		if !strings.HasPrefix(request.Model, "o1-mini") {
+		if !strings.HasPrefix(request.Model, "o1-mini") && !strings.HasPrefix(request.Model, "o1-preview") {
 			//修改第一个Message的内容，将system改为developer
 			if len(request.Messages) > 0 && request.Messages[0].Role == "system" {
 				request.Messages[0].Role = "developer"
@@ -236,11 +241,167 @@ func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInf
 }
 
 func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
+	switch info.RelayMode {
+	case constant.RelayModeImagesEdits:
+
+		var requestBody bytes.Buffer
+		writer := multipart.NewWriter(&requestBody)
+
+		writer.WriteField("model", request.Model)
+		// 获取所有表单字段
+		formData := c.Request.PostForm
+		// 遍历表单字段并打印输出
+		for key, values := range formData {
+			if key == "model" {
+				continue
+			}
+			for _, value := range values {
+				writer.WriteField(key, value)
+			}
+		}
+
+		// Parse the multipart form to handle both single image and multiple images
+		if err := c.Request.ParseMultipartForm(32 << 20); err != nil { // 32MB max memory
+			return nil, errors.New("failed to parse multipart form")
+		}
+
+		if c.Request.MultipartForm != nil && c.Request.MultipartForm.File != nil {
+			// Check if "image" field exists in any form, including array notation
+			var imageFiles []*multipart.FileHeader
+			var exists bool
+
+			// First check for standard "image" field
+			if imageFiles, exists = c.Request.MultipartForm.File["image"]; !exists || len(imageFiles) == 0 {
+				// If not found, check for "image[]" field
+				if imageFiles, exists = c.Request.MultipartForm.File["image[]"]; !exists || len(imageFiles) == 0 {
+					// If still not found, iterate through all fields to find any that start with "image["
+					foundArrayImages := false
+					for fieldName, files := range c.Request.MultipartForm.File {
+						if strings.HasPrefix(fieldName, "image[") && len(files) > 0 {
+							foundArrayImages = true
+							for _, file := range files {
+								imageFiles = append(imageFiles, file)
+							}
+						}
+					}
+
+					// If no image fields found at all
+					if !foundArrayImages && (len(imageFiles) == 0) {
+						return nil, errors.New("image is required")
+					}
+				}
+			}
+
+			// Process all image files
+			for i, fileHeader := range imageFiles {
+				file, err := fileHeader.Open()
+				if err != nil {
+					return nil, fmt.Errorf("failed to open image file %d: %w", i, err)
+				}
+				defer file.Close()
+
+				// If multiple images, use image[] as the field name
+				fieldName := "image"
+				if len(imageFiles) > 1 {
+					fieldName = "image[]"
+				}
+
+				// Determine MIME type based on file extension
+				mimeType := detectImageMimeType(fileHeader.Filename)
+
+				// Create a form file with the appropriate content type
+				h := make(textproto.MIMEHeader)
+				h.Set("Content-Disposition", fmt.Sprintf(`form-data; name="%s"; filename="%s"`, fieldName, fileHeader.Filename))
+				h.Set("Content-Type", mimeType)
+
+				part, err := writer.CreatePart(h)
+				if err != nil {
+					return nil, fmt.Errorf("create form part failed for image %d: %w", i, err)
+				}
+
+				if _, err := io.Copy(part, file); err != nil {
+					return nil, fmt.Errorf("copy file failed for image %d: %w", i, err)
+				}
+			}
+
+			// Handle mask file if present
+			if maskFiles, exists := c.Request.MultipartForm.File["mask"]; exists && len(maskFiles) > 0 {
+				maskFile, err := maskFiles[0].Open()
+				if err != nil {
+					return nil, errors.New("failed to open mask file")
+				}
+				defer maskFile.Close()
+
+				// Determine MIME type for mask file
+				mimeType := detectImageMimeType(maskFiles[0].Filename)
+
+				// Create a form file with the appropriate content type
+				h := make(textproto.MIMEHeader)
+				h.Set("Content-Disposition", fmt.Sprintf(`form-data; name="mask"; filename="%s"`, maskFiles[0].Filename))
+				h.Set("Content-Type", mimeType)
+
+				maskPart, err := writer.CreatePart(h)
+				if err != nil {
+					return nil, errors.New("create form file failed for mask")
+				}
+
+				if _, err := io.Copy(maskPart, maskFile); err != nil {
+					return nil, errors.New("copy mask file failed")
+				}
+			}
+		} else {
+			return nil, errors.New("no multipart form data found")
+		}
+
+		// 关闭 multipart 编写器以设置分界线
+		writer.Close()
+		c.Request.Header.Set("Content-Type", writer.FormDataContentType())
+		return bytes.NewReader(requestBody.Bytes()), nil
+
+	default:
+		return request, nil
+	}
+}
+
+// detectImageMimeType determines the MIME type based on the file extension
+func detectImageMimeType(filename string) string {
+	ext := strings.ToLower(filepath.Ext(filename))
+	switch ext {
+	case ".jpg", ".jpeg":
+		return "image/jpeg"
+	case ".png":
+		return "image/png"
+	case ".webp":
+		return "image/webp"
+	default:
+		// Try to detect from extension if possible
+		if strings.HasPrefix(ext, ".jp") {
+			return "image/jpeg"
+		}
+		// Default to png as a fallback
+		return "image/png"
+	}
+}
+
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// 模型后缀转换 reasoning effort
+	if strings.HasSuffix(request.Model, "-high") {
+		request.Reasoning.Effort = "high"
+		request.Model = strings.TrimSuffix(request.Model, "-high")
+	} else if strings.HasSuffix(request.Model, "-low") {
+		request.Reasoning.Effort = "low"
+		request.Model = strings.TrimSuffix(request.Model, "-low")
+	} else if strings.HasSuffix(request.Model, "-medium") {
+		request.Reasoning.Effort = "medium"
+		request.Model = strings.TrimSuffix(request.Model, "-medium")
+	}
 	return request, nil
 }
 
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
-	if info.RelayMode == constant.RelayModeAudioTranscription || info.RelayMode == constant.RelayModeAudioTranslation {
+	if info.RelayMode == constant.RelayModeAudioTranscription ||
+		info.RelayMode == constant.RelayModeAudioTranslation ||
+		info.RelayMode == constant.RelayModeImagesEdits {
 		return channel.DoFormRequest(a, c, info, requestBody)
 	} else if info.RelayMode == constant.RelayModeRealtime {
 		return channel.DoWssRequest(a, c, info, requestBody)
@@ -259,10 +420,16 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycom
 		fallthrough
 	case constant.RelayModeAudioTranscription:
 		err, usage = OpenaiSTTHandler(c, resp, info, a.ResponseFormat)
-	case constant.RelayModeImagesGenerations:
-		err, usage = OpenaiTTSHandler(c, resp, info)
+	case constant.RelayModeImagesGenerations, constant.RelayModeImagesEdits:
+		err, usage = OpenaiHandlerWithUsage(c, resp, info)
 	case constant.RelayModeRerank:
 		err, usage = common_handler.RerankHandler(c, info, resp)
+	case constant.RelayModeResponses:
+		if info.IsStream {
+			err, usage = OaiResponsesStreamHandler(c, resp, info)
+		} else {
+			err, usage = OaiResponsesHandler(c, resp, info)
+		}
 	default:
 		if info.IsStream {
 			err, usage = OaiStreamHandler(c, resp, info)

relay/channel/openai/helper.go

@@ -187,3 +187,10 @@ func handleFinalResponse(c *gin.Context, info *relaycommon.RelayInfo, lastStream
 		}
 	}
 }
+
+func sendResponsesStreamData(c *gin.Context, streamResponse dto.ResponsesStreamResponse, data string) {
+	if data == "" {
+		return
+	}
+	helper.ResponseChunkData(c, streamResponse, data)
+}

relay/channel/openai/relay-openai.go

@@ -215,10 +215,35 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayI
 			StatusCode: resp.StatusCode,
 		}, nil
 	}
+	
+	forceFormat := false
+	if forceFmt, ok := info.ChannelSetting[constant.ForceFormat].(bool); ok {
+		forceFormat = forceFmt
+	}
+
+	if simpleResponse.Usage.TotalTokens == 0 || (simpleResponse.Usage.PromptTokens == 0 && simpleResponse.Usage.CompletionTokens == 0) {
+		completionTokens := 0
+		for _, choice := range simpleResponse.Choices {
+			ctkm, _ := service.CountTextToken(choice.Message.StringContent()+choice.Message.ReasoningContent+choice.Message.Reasoning, info.UpstreamModelName)
+			completionTokens += ctkm
+		}
+		simpleResponse.Usage = dto.Usage{
+			PromptTokens:     info.PromptTokens,
+			CompletionTokens: completionTokens,
+			TotalTokens:      info.PromptTokens + completionTokens,
+		}
+	}
 
 	switch info.RelayFormat {
 	case relaycommon.RelayFormatOpenAI:
-		break
+		if forceFormat {
+			responseBody, err = json.Marshal(simpleResponse)
+			if err != nil {
+				return service.OpenAIErrorWrapper(err, "marshal_response_body_failed", http.StatusInternalServerError), nil
+			}
+		} else {
+			break
+		}
 	case relaycommon.RelayFormatClaude:
 		claudeResp := service.ResponseOpenAI2Claude(&simpleResponse, info)
 		claudeRespStr, err := json.Marshal(claudeResp)
@@ -244,18 +269,6 @@ func OpenaiHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayI
 		common.SysError("error copying response body: " + err.Error())
 	}
 	resp.Body.Close()
-	if simpleResponse.Usage.TotalTokens == 0 || (simpleResponse.Usage.PromptTokens == 0 && simpleResponse.Usage.CompletionTokens == 0) {
-		completionTokens := 0
-		for _, choice := range simpleResponse.Choices {
-			ctkm, _ := service.CountTextToken(choice.Message.StringContent()+choice.Message.ReasoningContent+choice.Message.Reasoning, info.UpstreamModelName)
-			completionTokens += ctkm
-		}
-		simpleResponse.Usage = dto.Usage{
-			PromptTokens:     info.PromptTokens,
-			CompletionTokens: completionTokens,
-			TotalTokens:      info.PromptTokens + completionTokens,
-		}
-	}
 	return nil, &simpleResponse.Usage
 }
 
@@ -595,3 +608,52 @@ func preConsumeUsage(ctx *gin.Context, info *relaycommon.RelayInfo, usage *dto.R
 	err := service.PreWssConsumeQuota(ctx, info, usage)
 	return err
 }
+
+func OpenaiHandlerWithUsage(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	// Reset response body
+	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	// We shouldn't set the header before we parse the response body, because the parse part may fail.
+	// And then we will have to send an error response, but in this case, the header has already been set.
+	// So the httpClient will be confused by the response.
+	// For example, Postman will report error, and we cannot check the response at all.
+	for k, v := range resp.Header {
+		c.Writer.Header().Set(k, v[0])
+	}
+	// reset content length
+	c.Writer.Header().Set("Content-Length", fmt.Sprintf("%d", len(responseBody)))
+	c.Writer.WriteHeader(resp.StatusCode)
+	_, err = io.Copy(c.Writer, resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "copy_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+
+	var usageResp dto.SimpleResponse
+	err = json.Unmarshal(responseBody, &usageResp)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "parse_response_body_failed", http.StatusInternalServerError), nil
+	}
+	// format
+	if usageResp.InputTokens > 0 {
+		usageResp.PromptTokens += usageResp.InputTokens
+	}
+	if usageResp.OutputTokens > 0 {
+		usageResp.CompletionTokens += usageResp.OutputTokens
+	}
+	if usageResp.InputTokensDetails != nil {
+		usageResp.PromptTokensDetails.ImageTokens += usageResp.InputTokensDetails.ImageTokens
+		usageResp.PromptTokensDetails.TextTokens += usageResp.InputTokensDetails.TextTokens
+	}
+	return nil, &usageResp.Usage
+}

relay/channel/openai/relay_responses.go

@@ -0,0 +1,119 @@
+package openai
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"net/http"
+	"one-api/common"
+	"one-api/dto"
+	relaycommon "one-api/relay/common"
+	"one-api/relay/helper"
+	"one-api/service"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+)
+
+func OaiResponsesHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	// read response body
+	var responsesResponse dto.OpenAIResponsesResponse
+	responseBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = resp.Body.Close()
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "close_response_body_failed", http.StatusInternalServerError), nil
+	}
+	err = common.DecodeJson(responseBody, &responsesResponse)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "unmarshal_response_body_failed", http.StatusInternalServerError), nil
+	}
+	if responsesResponse.Error != nil {
+		return &dto.OpenAIErrorWithStatusCode{
+			Error: dto.OpenAIError{
+				Message: responsesResponse.Error.Message,
+				Type:    "openai_error",
+				Code:    responsesResponse.Error.Code,
+			},
+			StatusCode: resp.StatusCode,
+		}, nil
+	}
+
+	// reset response body
+	resp.Body = io.NopCloser(bytes.NewBuffer(responseBody))
+	// We shouldn't set the header before we parse the response body, because the parse part may fail.
+	// And then we will have to send an error response, but in this case, the header has already been set.
+	// So the httpClient will be confused by the response.
+	// For example, Postman will report error, and we cannot check the response at all.
+	for k, v := range resp.Header {
+		c.Writer.Header().Set(k, v[0])
+	}
+	c.Writer.WriteHeader(resp.StatusCode)
+	// copy response body
+	_, err = io.Copy(c.Writer, resp.Body)
+	if err != nil {
+		common.SysError("error copying response body: " + err.Error())
+	}
+	resp.Body.Close()
+	// compute usage
+	usage := dto.Usage{}
+	usage.PromptTokens = responsesResponse.Usage.InputTokens
+	usage.CompletionTokens = responsesResponse.Usage.OutputTokens
+	usage.TotalTokens = responsesResponse.Usage.TotalTokens
+	// 解析 Tools 用量
+	for _, tool := range responsesResponse.Tools {
+		info.ResponsesUsageInfo.BuiltInTools[tool.Type].CallCount++
+	}
+	return nil, &usage
+}
+
+func OaiResponsesStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (*dto.OpenAIErrorWithStatusCode, *dto.Usage) {
+	if resp == nil || resp.Body == nil {
+		common.LogError(c, "invalid response or response body")
+		return service.OpenAIErrorWrapper(fmt.Errorf("invalid response"), "invalid_response", http.StatusInternalServerError), nil
+	}
+
+	var usage = &dto.Usage{}
+	var responseTextBuilder strings.Builder
+
+	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+
+		// 检查当前数据是否包含 completed 状态和 usage 信息
+		var streamResponse dto.ResponsesStreamResponse
+		if err := common.DecodeJsonStr(data, &streamResponse); err == nil {
+			sendResponsesStreamData(c, streamResponse, data)
+			switch streamResponse.Type {
+			case "response.completed":
+				usage.PromptTokens = streamResponse.Response.Usage.InputTokens
+				usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
+				usage.TotalTokens = streamResponse.Response.Usage.TotalTokens
+			case "response.output_text.delta":
+				// 处理输出文本
+				responseTextBuilder.WriteString(streamResponse.Delta)
+			case dto.ResponsesOutputTypeItemDone:
+				// 函数调用处理
+				if streamResponse.Item != nil {
+					switch streamResponse.Item.Type {
+					case dto.BuildInCallWebSearchCall:
+						info.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview].CallCount++
+					}
+				}
+			}
+		}
+		return true
+	})
+
+	if usage.CompletionTokens == 0 {
+		// 计算输出文本的 token 数量
+		tempStr := responseTextBuilder.String()
+		if len(tempStr) > 0 {
+			// 非正常结束，使用输出文本的 token 数量
+			completionTokens, _ := service.CountTextToken(tempStr, info.UpstreamModelName)
+			usage.CompletionTokens = completionTokens
+		}
+	}
+
+	return nil, usage
+}

relay/channel/palm/adaptor.go

@@ -3,13 +3,14 @@ package palm
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	relaycommon "one-api/relay/common"
 	"one-api/service"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -60,6 +61,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/perplexity/adaptor.go

@@ -3,13 +3,14 @@ package perplexity
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -63,6 +64,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/siliconflow/adaptor.go

@@ -3,7 +3,6 @@ package siliconflow
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,8 @@ import (
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -58,6 +59,11 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	return request, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/tencent/adaptor.go

@@ -3,7 +3,6 @@ package tencent
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/common"
@@ -13,6 +12,8 @@ import (
 	"one-api/service"
 	"strconv"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -84,6 +85,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/vertex/adaptor.go

@@ -4,7 +4,6 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -13,7 +12,10 @@ import (
 	"one-api/relay/channel/gemini"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
+	"one-api/setting/model_setting"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 const (
@@ -77,6 +79,15 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	a.AccountCredentials = *adc
 	suffix := ""
 	if a.RequestMode == RequestModeGemini {
+		if model_setting.GetGeminiSettings().ThinkingAdapterEnabled {
+			// suffix -thinking and -nothinking
+			if strings.HasSuffix(info.OriginModelName, "-thinking") {
+				info.UpstreamModelName = strings.TrimSuffix(info.UpstreamModelName, "-thinking")
+			} else if strings.HasSuffix(info.OriginModelName, "-nothinking") {
+				info.UpstreamModelName = strings.TrimSuffix(info.UpstreamModelName, "-nothinking")
+			}
+		}
+
 		if info.IsStream {
 			suffix = "streamGenerateContent?alt=sse"
 		} else {
@@ -164,6 +175,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/volcengine/adaptor.go

@@ -3,7 +3,6 @@ package volcengine
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -12,6 +11,8 @@ import (
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -71,6 +72,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return request, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/channel/xai/adaptor.go

@@ -2,14 +2,17 @@ package xai
 
 import (
 	"errors"
-	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
+	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	"strings"
+
+	"one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -27,15 +30,20 @@ func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInf
 }
 
 func (a *Adaptor) ConvertImageRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.ImageRequest) (any, error) {
-	request.Size = ""
-	return request, nil
+	xaiRequest := ImageRequest{
+		Model:          request.Model,
+		Prompt:         request.Prompt,
+		N:              request.N,
+		ResponseFormat: request.ResponseFormat,
+	}
+	return xaiRequest, nil
 }
 
 func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
-	return fmt.Sprintf("%s/v1/chat/completions", info.BaseUrl), nil
+	return relaycommon.GetFullRequestURL(info.BaseUrl, info.RequestURLPath, info.ChannelType), nil
 }
 
 func (a *Adaptor) SetupRequestHeader(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) error {
@@ -78,20 +86,26 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not available")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
-	if info.IsStream {
-		err, usage = xAIStreamHandler(c, resp, info)
-	} else {
-		err, usage = xAIHandler(c, resp, info)
+	switch info.RelayMode {
+	case constant.RelayModeImagesGenerations, constant.RelayModeImagesEdits:
+		err, usage = openai.OpenaiHandlerWithUsage(c, resp, info)
+	default:
+		if info.IsStream {
+			err, usage = xAIStreamHandler(c, resp, info)
+		} else {
+			err, usage = xAIHandler(c, resp, info)
+		}
 	}
-	//if _, ok := usage.(*dto.Usage); ok && usage != nil {
-	//	usage.(*dto.Usage).CompletionTokens = usage.(*dto.Usage).TotalTokens - usage.(*dto.Usage).PromptTokens
-	//}
-
 	return
 }
 

relay/channel/xai/dto.go

@@ -12,3 +12,16 @@ type ChatCompletionResponse struct {
 	Usage             *dto.Usage `json:"usage"`
 	SystemFingerprint string     `json:"system_fingerprint"`
 }
+
+// quality, size or style are not supported by xAI API at the moment.
+type ImageRequest struct {
+	Model          string          `json:"model"`
+	Prompt         string          `json:"prompt" binding:"required"`
+	N              int             `json:"n,omitempty"`
+	// Size           string          `json:"size,omitempty"`
+	// Quality        string          `json:"quality,omitempty"`
+	ResponseFormat string          `json:"response_format,omitempty"`
+	// Style          string          `json:"style,omitempty"`
+	// User           string          `json:"user,omitempty"`
+	// ExtraFields    json.RawMessage `json:"extra_fields,omitempty"`
+}

relay/channel/xunfei/adaptor.go

@@ -2,7 +2,6 @@ package xunfei
 
 import (
 	"errors"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -10,6 +9,8 @@ import (
 	relaycommon "one-api/relay/common"
 	"one-api/service"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -61,6 +62,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return nil, errors.New("not implemented")
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	// xunfei's request is not http request, so we don't need to do anything here
 	dummyResp := &http.Response{}

relay/channel/zhipu/adaptor.go

@@ -3,12 +3,13 @@ package zhipu
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
 	relaycommon "one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -71,6 +72,11 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 	return channel.DoApiRequest(a, c, info, requestBody)
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *dto.OpenAIErrorWithStatusCode) {
 	if info.IsStream {
 		err, usage = zhipuStreamHandler(c, resp)

relay/channel/zhipu_4v/adaptor.go

@@ -3,7 +3,6 @@ package zhipu_4v
 import (
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
@@ -11,6 +10,8 @@ import (
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	relayconstant "one-api/relay/constant"
+
+	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -70,6 +71,11 @@ func (a *Adaptor) ConvertEmbeddingRequest(c *gin.Context, info *relaycommon.Rela
 	return request, nil
 }
 
+func (a *Adaptor) ConvertOpenAIResponsesRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.OpenAIResponsesRequest) (any, error) {
+	// TODO implement me
+	return nil, errors.New("not implemented")
+}
+
 func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (any, error) {
 	return channel.DoApiRequest(a, c, info, requestBody)
 }

relay/common/relay_info.go

@@ -36,6 +36,7 @@ type ClaudeConvertInfo struct {
 const (
 	RelayFormatOpenAI = "openai"
 	RelayFormatClaude = "claude"
+	RelayFormatGemini = "gemini"
 )
 
 type RerankerInfo struct {
@@ -43,6 +44,16 @@ type RerankerInfo struct {
 	ReturnDocuments bool
 }
 
+type BuildInToolInfo struct {
+	ToolName          string
+	CallCount         int
+	SearchContextSize string
+}
+
+type ResponsesUsageInfo struct {
+	BuiltInTools map[string]*BuildInToolInfo
+}
+
 type RelayInfo struct {
 	ChannelType       int
 	ChannelId         int
@@ -87,9 +98,11 @@ type RelayInfo struct {
 	UserQuota            int
 	RelayFormat          string
 	SendResponseCount    int
+	ChannelCreateTime    int64
 	ThinkingContentInfo
 	*ClaudeConvertInfo
 	*RerankerInfo
+	*ResponsesUsageInfo
 }
 
 // 定义支持流式选项的通道类型
@@ -103,6 +116,8 @@ var streamSupportedChannels = map[int]bool{
 	common.ChannelTypeVolcEngine: true,
 	common.ChannelTypeOllama:     true,
 	common.ChannelTypeXai:        true,
+	common.ChannelTypeDeepSeek:   true,
+	common.ChannelTypeBaiduV2:    true,
 }
 
 func GenRelayInfoWs(c *gin.Context, ws *websocket.Conn) *RelayInfo {
@@ -134,6 +149,31 @@ func GenRelayInfoRerank(c *gin.Context, req *dto.RerankRequest) *RelayInfo {
 	return info
 }
 
+func GenRelayInfoResponses(c *gin.Context, req *dto.OpenAIResponsesRequest) *RelayInfo {
+	info := GenRelayInfo(c)
+	info.RelayMode = relayconstant.RelayModeResponses
+	info.ResponsesUsageInfo = &ResponsesUsageInfo{
+		BuiltInTools: make(map[string]*BuildInToolInfo),
+	}
+	if len(req.Tools) > 0 {
+		for _, tool := range req.Tools {
+			info.ResponsesUsageInfo.BuiltInTools[tool.Type] = &BuildInToolInfo{
+				ToolName:  tool.Type,
+				CallCount: 0,
+			}
+			switch tool.Type {
+			case dto.BuildInToolWebSearchPreview:
+				if tool.SearchContextSize == "" {
+					tool.SearchContextSize = "medium"
+				}
+				info.ResponsesUsageInfo.BuiltInTools[tool.Type].SearchContextSize = tool.SearchContextSize
+			}
+		}
+	}
+	info.IsStream = req.Stream
+	return info
+}
+
 func GenRelayInfo(c *gin.Context) *RelayInfo {
 	channelType := c.GetInt("channel_type")
 	channelId := c.GetInt("channel_id")
@@ -170,14 +210,15 @@ func GenRelayInfo(c *gin.Context) *RelayInfo {
 		OriginModelName:   c.GetString("original_model"),
 		UpstreamModelName: c.GetString("original_model"),
 		//RecodeModelName:   c.GetString("original_model"),
-		IsModelMapped:  false,
-		ApiType:        apiType,
-		ApiVersion:     c.GetString("api_version"),
-		ApiKey:         strings.TrimPrefix(c.Request.Header.Get("Authorization"), "Bearer "),
-		Organization:   c.GetString("channel_organization"),
-		ChannelSetting: channelSetting,
-		ParamOverride:  paramOverride,
-		RelayFormat:    RelayFormatOpenAI,
+		IsModelMapped:     false,
+		ApiType:           apiType,
+		ApiVersion:        c.GetString("api_version"),
+		ApiKey:            strings.TrimPrefix(c.Request.Header.Get("Authorization"), "Bearer "),
+		Organization:      c.GetString("channel_organization"),
+		ChannelSetting:    channelSetting,
+		ChannelCreateTime: c.GetInt64("channel_create_time"),
+		ParamOverride:     paramOverride,
+		RelayFormat:       RelayFormatOpenAI,
 		ThinkingContentInfo: ThinkingContentInfo{
 			IsFirstThinkingContent:  true,
 			SendLastThinkingContent: false,
@@ -200,6 +241,10 @@ func GenRelayInfo(c *gin.Context) *RelayInfo {
 	if streamSupportedChannels[info.ChannelType] {
 		info.SupportStreamOptions = true
 	}
+	// responses 模式不支持 StreamOptions
+	if relayconstant.RelayModeResponses == info.RelayMode {
+		info.SupportStreamOptions = false
+	}
 	return info
 }
 

relay/constant/relay_mode.go

@@ -12,6 +12,7 @@ const (
 	RelayModeEmbeddings
 	RelayModeModerations
 	RelayModeImagesGenerations
+	RelayModeImagesEdits
 	RelayModeEdits
 
 	RelayModeMidjourneyImagine
@@ -39,6 +40,8 @@ const (
 
 	RelayModeRerank
 
+	RelayModeResponses
+
 	RelayModeRealtime
 )
 
@@ -56,8 +59,12 @@ func Path2RelayMode(path string) int {
 		relayMode = RelayModeModerations
 	} else if strings.HasPrefix(path, "/v1/images/generations") {
 		relayMode = RelayModeImagesGenerations
+	} else if strings.HasPrefix(path, "/v1/images/edits") {
+		relayMode = RelayModeImagesEdits
 	} else if strings.HasPrefix(path, "/v1/edits") {
 		relayMode = RelayModeEdits
+	} else if strings.HasPrefix(path, "/v1/responses") {
+		relayMode = RelayModeResponses
 	} else if strings.HasPrefix(path, "/v1/audio/speech") {
 		relayMode = RelayModeAudioSpeech
 	} else if strings.HasPrefix(path, "/v1/audio/transcriptions") {

relay/helper/common.go

@@ -12,11 +12,19 @@ import (
 )
 
 func SetEventStreamHeaders(c *gin.Context) {
-	c.Writer.Header().Set("Content-Type", "text/event-stream")
-	c.Writer.Header().Set("Cache-Control", "no-cache")
-	c.Writer.Header().Set("Connection", "keep-alive")
-	c.Writer.Header().Set("Transfer-Encoding", "chunked")
-	c.Writer.Header().Set("X-Accel-Buffering", "no")
+    // 检查是否已经设置过头部
+    if _, exists := c.Get("event_stream_headers_set"); exists {
+        return
+    }
+    
+    c.Writer.Header().Set("Content-Type", "text/event-stream")
+    c.Writer.Header().Set("Cache-Control", "no-cache")
+    c.Writer.Header().Set("Connection", "keep-alive")
+    c.Writer.Header().Set("Transfer-Encoding", "chunked")
+    c.Writer.Header().Set("X-Accel-Buffering", "no")
+    
+    // 设置标志，表示头部已经设置过
+    c.Set("event_stream_headers_set", true)
 }
 
 func ClaudeData(c *gin.Context, resp dto.ClaudeResponse) error {
@@ -43,6 +51,14 @@ func ClaudeChunkData(c *gin.Context, resp dto.ClaudeResponse, data string) {
 	}
 }
 
+func ResponseChunkData(c *gin.Context, resp dto.ResponsesStreamResponse, data string) {
+	c.Render(-1, common.CustomEvent{Data: fmt.Sprintf("event: %s\n", resp.Type)})
+	c.Render(-1, common.CustomEvent{Data: fmt.Sprintf("data: %s", data)})
+	if flusher, ok := c.Writer.(http.Flusher); ok {
+		flusher.Flush()
+	}
+}
+
 func StringData(c *gin.Context, str string) error {
 	//str = strings.TrimPrefix(str, "data: ")
 	//str = strings.TrimSuffix(str, "\r")

relay/helper/model_mapped.go

@@ -2,9 +2,11 @@ package helper
 
 import (
 	"encoding/json"
+	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"one-api/relay/common"
+
+	"github.com/gin-gonic/gin"
 )
 
 func ModelMappedHelper(c *gin.Context, info *common.RelayInfo) error {
@@ -16,9 +18,36 @@ func ModelMappedHelper(c *gin.Context, info *common.RelayInfo) error {
 		if err != nil {
 			return fmt.Errorf("unmarshal_model_mapping_failed")
 		}
-		if modelMap[info.OriginModelName] != "" {
-			info.UpstreamModelName = modelMap[info.OriginModelName]
-			info.IsModelMapped = true
+
+		// 支持链式模型重定向，最终使用链尾的模型
+		currentModel := info.OriginModelName
+		visitedModels := map[string]bool{
+			currentModel: true,
+		}
+		for {
+			if mappedModel, exists := modelMap[currentModel]; exists && mappedModel != "" {
+				// 模型重定向循环检测，避免无限循环
+				if visitedModels[mappedModel] {
+					if mappedModel == currentModel {
+						if currentModel == info.OriginModelName {
+							info.IsModelMapped = false
+							return nil
+						} else {
+							info.IsModelMapped = true
+							break
+						}
+					}
+					return errors.New("model_mapping_contains_cycle")
+				}
+				visitedModels[mappedModel] = true
+				currentModel = mappedModel
+				info.IsModelMapped = true
+			} else {
+				break
+			}
+		}
+		if info.IsModelMapped {
+			info.UpstreamModelName = currentModel
 		}
 	}
 	return nil

relay/helper/price.go

@@ -15,14 +15,15 @@ type PriceData struct {
 	ModelRatio             float64
 	CompletionRatio        float64
 	CacheRatio             float64
+	CacheCreationRatio     float64
+	ImageRatio             float64
 	GroupRatio             float64
 	UsePrice               bool
-	CacheCreationRatio     float64
 	ShouldPreConsumedQuota int
 }
 
 func (p PriceData) ToSetting() string {
-	return fmt.Sprintf("ModelPrice: %f, ModelRatio: %f, CompletionRatio: %f, CacheRatio: %f, GroupRatio: %f, UsePrice: %t, CacheCreationRatio: %f, ShouldPreConsumedQuota: %d", p.ModelPrice, p.ModelRatio, p.CompletionRatio, p.CacheRatio, p.GroupRatio, p.UsePrice, p.CacheCreationRatio, p.ShouldPreConsumedQuota)
+	return fmt.Sprintf("ModelPrice: %f, ModelRatio: %f, CompletionRatio: %f, CacheRatio: %f, GroupRatio: %f, UsePrice: %t, CacheCreationRatio: %f, ShouldPreConsumedQuota: %d, ImageRatio: %f", p.ModelPrice, p.ModelRatio, p.CompletionRatio, p.CacheRatio, p.GroupRatio, p.UsePrice, p.CacheCreationRatio, p.ShouldPreConsumedQuota, p.ImageRatio)
 }
 
 func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens int, maxTokens int) (PriceData, error) {
@@ -32,6 +33,7 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
 	var modelRatio float64
 	var completionRatio float64
 	var cacheRatio float64
+	var imageRatio float64
 	var cacheCreationRatio float64
 	if !usePrice {
 		preConsumedTokens := common.PreConsumedQuota
@@ -55,6 +57,7 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
 		completionRatio = operation_setting.GetCompletionRatio(info.OriginModelName)
 		cacheRatio, _ = operation_setting.GetCacheRatio(info.OriginModelName)
 		cacheCreationRatio, _ = operation_setting.GetCreateCacheRatio(info.OriginModelName)
+		imageRatio, _ = operation_setting.GetImageRatio(info.OriginModelName)
 		ratio := modelRatio * groupRatio
 		preConsumedQuota = int(float64(preConsumedTokens) * ratio)
 	} else {
@@ -68,6 +71,7 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
 		GroupRatio:             groupRatio,
 		UsePrice:               usePrice,
 		CacheRatio:             cacheRatio,
+		ImageRatio:             imageRatio,
 		CacheCreationRatio:     cacheCreationRatio,
 		ShouldPreConsumedQuota: preConsumedQuota,
 	}

relay/helper/stream_scanner.go

@@ -3,7 +3,6 @@ package helper
 import (
 	"bufio"
 	"context"
-	"github.com/bytedance/gopkg/util/gopool"
 	"io"
 	"net/http"
 	"one-api/common"
@@ -14,6 +13,8 @@ import (
 	"sync"
 	"time"
 
+	"github.com/bytedance/gopkg/util/gopool"
+
 	"github.com/gin-gonic/gin"
 )
 
@@ -32,7 +33,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 	defer resp.Body.Close()
 
 	streamingTimeout := time.Duration(constant.StreamingTimeout) * time.Second
-	if strings.HasPrefix(info.UpstreamModelName, "o1") || strings.HasPrefix(info.UpstreamModelName, "o3") {
+	if strings.HasPrefix(info.UpstreamModelName, "o") {
 		// twice timeout for thinking model
 		streamingTimeout *= 2
 	}
@@ -115,7 +116,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 			}
 			data = data[5:]
 			data = strings.TrimLeft(data, " ")
-			data = strings.TrimSuffix(data, "\"")
+			data = strings.TrimSuffix(data, "\r")
 			if !strings.HasPrefix(data, "[DONE]") {
 				info.SetFirstResponseTime()
 				writeMutex.Lock() // Lock before writing

relay/relay-image.go

@@ -5,21 +5,83 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
-	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/common"
 	"one-api/dto"
 	"one-api/model"
 	relaycommon "one-api/relay/common"
+	relayconstant "one-api/relay/constant"
 	"one-api/relay/helper"
 	"one-api/service"
 	"one-api/setting"
 	"strings"
+
+	"github.com/gin-gonic/gin"
 )
 
 func getAndValidImageRequest(c *gin.Context, info *relaycommon.RelayInfo) (*dto.ImageRequest, error) {
 	imageRequest := &dto.ImageRequest{}
+
+	switch info.RelayMode {
+	case relayconstant.RelayModeImagesEdits:
+		_, err := c.MultipartForm()
+		if err != nil {
+			return nil, err
+		}
+		formData := c.Request.PostForm
+		imageRequest.Prompt = formData.Get("prompt")
+		imageRequest.Model = formData.Get("model")
+		imageRequest.N = common.String2Int(formData.Get("n"))
+		imageRequest.Quality = formData.Get("quality")
+		imageRequest.Size = formData.Get("size")
+
+		if imageRequest.Model == "gpt-image-1" {
+			if imageRequest.Quality == "" {
+				imageRequest.Quality = "standard"
+			}
+		}
+	default:
+		err := common.UnmarshalBodyReusable(c, imageRequest)
+		if err != nil {
+			return nil, err
+		}
+		// Not "256x256", "512x512", or "1024x1024"
+		if imageRequest.Model == "dall-e-2" || imageRequest.Model == "dall-e" {
+			if imageRequest.Size != "" && imageRequest.Size != "256x256" && imageRequest.Size != "512x512" && imageRequest.Size != "1024x1024" {
+				return nil, errors.New("size must be one of 256x256, 512x512, or 1024x1024 for dall-e-2 or dall-e")
+			}
+		} else if imageRequest.Model == "dall-e-3" {
+			if imageRequest.Size != "" && imageRequest.Size != "1024x1024" && imageRequest.Size != "1024x1792" && imageRequest.Size != "1792x1024" {
+				return nil, errors.New("size must be one of 1024x1024, 1024x1792 or 1792x1024 for dall-e-3")
+			}
+			if imageRequest.Quality == "" {
+				imageRequest.Quality = "standard"
+			}
+			// N should between 1 and 10
+			//if imageRequest.N != 0 && (imageRequest.N < 1 || imageRequest.N > 10) {
+			//	return service.OpenAIErrorWrapper(errors.New("n must be between 1 and 10"), "invalid_field_value", http.StatusBadRequest)
+			//}
+		}
+	}
+
+	if imageRequest.Prompt == "" {
+		return nil, errors.New("prompt is required")
+	}
+
+	if imageRequest.Model == "" {
+		imageRequest.Model = "dall-e-2"
+	}
+	if strings.Contains(imageRequest.Size, "×") {
+		return nil, errors.New("size an unexpected error occurred in the parameter, please use 'x' instead of the multiplication sign '×'")
+	}
+	if imageRequest.N == 0 {
+		imageRequest.N = 1
+	}
+	if imageRequest.Size == "" {
+		imageRequest.Size = "1024x1024"
+	}
+
 	err := common.UnmarshalBodyReusable(c, imageRequest)
 	if err != nil {
 		return nil, err
@@ -39,6 +101,10 @@ func getAndValidImageRequest(c *gin.Context, info *relaycommon.RelayInfo) (*dto.
 	if imageRequest.Model == "" {
 		imageRequest.Model = "dall-e-2"
 	}
+	// x.ai grok-2-image not support size, quality or style
+	if imageRequest.Size == "empty" {
+		imageRequest.Size = ""
+	}
 
 	// Not "256x256", "512x512", or "1024x1024"
 	if imageRequest.Model == "dall-e-2" || imageRequest.Model == "dall-e" {
@@ -86,43 +152,59 @@ func ImageHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 
 	imageRequest.Model = relayInfo.UpstreamModelName
 
-	priceData, err := helper.ModelPriceHelper(c, relayInfo, 0, 0)
+	priceData, err := helper.ModelPriceHelper(c, relayInfo, len(imageRequest.Prompt), 0)
 	if err != nil {
 		return service.OpenAIErrorWrapperLocal(err, "model_price_error", http.StatusInternalServerError)
 	}
+	var preConsumedQuota int
+	var quota int
+	var userQuota int
 	if !priceData.UsePrice {
 		// modelRatio 16 = modelPrice $0.04
 		// per 1 modelRatio = $0.04 / 16
-		priceData.ModelPrice = 0.0025 * priceData.ModelRatio
-	}
-
-	userQuota, err := model.GetUserQuota(relayInfo.UserId, false)
-
-	sizeRatio := 1.0
-	// Size
-	if imageRequest.Size == "256x256" {
-		sizeRatio = 0.4
-	} else if imageRequest.Size == "512x512" {
-		sizeRatio = 0.45
-	} else if imageRequest.Size == "1024x1024" {
-		sizeRatio = 1
-	} else if imageRequest.Size == "1024x1792" || imageRequest.Size == "1792x1024" {
-		sizeRatio = 2
-	}
-
-	qualityRatio := 1.0
-	if imageRequest.Model == "dall-e-3" && imageRequest.Quality == "hd" {
-		qualityRatio = 2.0
-		if imageRequest.Size == "1024x1792" || imageRequest.Size == "1792x1024" {
-			qualityRatio = 1.5
+		// priceData.ModelPrice = 0.0025 * priceData.ModelRatio
+		var openaiErr *dto.OpenAIErrorWithStatusCode
+		preConsumedQuota, userQuota, openaiErr = preConsumeQuota(c, priceData.ShouldPreConsumedQuota, relayInfo)
+		if openaiErr != nil {
+			return openaiErr
 		}
-	}
+		defer func() {
+			if openaiErr != nil {
+				returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
+			}
+		}()
 
-	priceData.ModelPrice *= sizeRatio * qualityRatio * float64(imageRequest.N)
-	quota := int(priceData.ModelPrice * priceData.GroupRatio * common.QuotaPerUnit)
+	} else {
+		sizeRatio := 1.0
+		// Size
+		if imageRequest.Size == "256x256" {
+			sizeRatio = 0.4
+		} else if imageRequest.Size == "512x512" {
+			sizeRatio = 0.45
+		} else if imageRequest.Size == "1024x1024" {
+			sizeRatio = 1
+		} else if imageRequest.Size == "1024x1792" || imageRequest.Size == "1792x1024" {
+			sizeRatio = 2
+		}
 
-	if userQuota-quota < 0 {
-		return service.OpenAIErrorWrapperLocal(fmt.Errorf("image pre-consumed quota failed, user quota: %s, need quota: %s", common.FormatQuota(userQuota), common.FormatQuota(quota)), "insufficient_user_quota", http.StatusForbidden)
+		qualityRatio := 1.0
+		if imageRequest.Model == "dall-e-3" && imageRequest.Quality == "hd" {
+			qualityRatio = 2.0
+			if imageRequest.Size == "1024x1792" || imageRequest.Size == "1792x1024" {
+				qualityRatio = 1.5
+			}
+		}
+
+		// reset model price
+		priceData.ModelPrice *= sizeRatio * qualityRatio * float64(imageRequest.N)
+		quota = int(priceData.ModelPrice * priceData.GroupRatio * common.QuotaPerUnit)
+		userQuota, err = model.GetUserQuota(relayInfo.UserId, false)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "get_user_quota_failed", http.StatusInternalServerError)
+		}
+		if userQuota-quota < 0 {
+			return service.OpenAIErrorWrapperLocal(fmt.Errorf("image pre-consumed quota failed, user quota: %s, need quota: %s", common.FormatQuota(userQuota), common.FormatQuota(quota)), "insufficient_user_quota", http.StatusForbidden)
+		}
 	}
 
 	adaptor := GetAdaptor(relayInfo.ApiType)
@@ -137,12 +219,15 @@ func ImageHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 	if err != nil {
 		return service.OpenAIErrorWrapperLocal(err, "convert_request_failed", http.StatusInternalServerError)
 	}
-
-	jsonData, err := json.Marshal(convertedRequest)
-	if err != nil {
-		return service.OpenAIErrorWrapperLocal(err, "json_marshal_failed", http.StatusInternalServerError)
+	if relayInfo.RelayMode == relayconstant.RelayModeImagesEdits {
+		requestBody = convertedRequest.(io.Reader)
+	} else {
+		jsonData, err := json.Marshal(convertedRequest)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "json_marshal_failed", http.StatusInternalServerError)
+		}
+		requestBody = bytes.NewBuffer(jsonData)
 	}
-	requestBody = bytes.NewBuffer(jsonData)
 
 	statusCodeMappingStr := c.GetString("status_code_mapping")
 
@@ -162,24 +247,25 @@ func ImageHelper(c *gin.Context) *dto.OpenAIErrorWithStatusCode {
 		}
 	}
 
-	_, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
+	usage, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
 	if openaiErr != nil {
 		// reset status code 重置状态码
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 		return openaiErr
 	}
 
-	usage := &dto.Usage{
-		PromptTokens: imageRequest.N,
-		TotalTokens:  imageRequest.N,
+	if usage.(*dto.Usage).TotalTokens == 0 {
+		usage.(*dto.Usage).TotalTokens = imageRequest.N
+	}
+	if usage.(*dto.Usage).PromptTokens == 0 {
+		usage.(*dto.Usage).PromptTokens = imageRequest.N
 	}
-
 	quality := "standard"
 	if imageRequest.Quality == "hd" {
 		quality = "hd"
 	}
 
 	logContent := fmt.Sprintf("大小 %s, 品质 %s", imageRequest.Size, quality)
-	postConsumeQuota(c, relayInfo, usage, 0, userQuota, priceData, logContent)
+	postConsumeQuota(c, relayInfo, usage.(*dto.Usage), preConsumedQuota, userQuota, priceData, logContent)
 	return nil
 }

relay/relay-responses.go

@@ -0,0 +1,171 @@
+package relay
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"one-api/common"
+	"one-api/dto"
+	relaycommon "one-api/relay/common"
+	"one-api/relay/helper"
+	"one-api/service"
+	"one-api/setting"
+	"one-api/setting/model_setting"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+)
+
+func getAndValidateResponsesRequest(c *gin.Context) (*dto.OpenAIResponsesRequest, error) {
+	request := &dto.OpenAIResponsesRequest{}
+	err := common.UnmarshalBodyReusable(c, request)
+	if err != nil {
+		return nil, err
+	}
+	if request.Model == "" {
+		return nil, errors.New("model is required")
+	}
+	if len(request.Input) == 0 {
+		return nil, errors.New("input is required")
+	}
+	return request, nil
+
+}
+
+func checkInputSensitive(textRequest *dto.OpenAIResponsesRequest, info *relaycommon.RelayInfo) ([]string, error) {
+	sensitiveWords, err := service.CheckSensitiveInput(textRequest.Input)
+	return sensitiveWords, err
+}
+
+func getInputTokens(req *dto.OpenAIResponsesRequest, info *relaycommon.RelayInfo) (int, error) {
+	inputTokens, err := service.CountTokenInput(req.Input, req.Model)
+	info.PromptTokens = inputTokens
+	return inputTokens, err
+}
+
+func ResponsesHelper(c *gin.Context) (openaiErr *dto.OpenAIErrorWithStatusCode) {
+	req, err := getAndValidateResponsesRequest(c)
+	if err != nil {
+		common.LogError(c, fmt.Sprintf("getAndValidateResponsesRequest error: %s", err.Error()))
+		return service.OpenAIErrorWrapperLocal(err, "invalid_responses_request", http.StatusBadRequest)
+	}
+
+	relayInfo := relaycommon.GenRelayInfoResponses(c, req)
+
+	if setting.ShouldCheckPromptSensitive() {
+		sensitiveWords, err := checkInputSensitive(req, relayInfo)
+		if err != nil {
+			common.LogWarn(c, fmt.Sprintf("user sensitive words detected: %s", strings.Join(sensitiveWords, ", ")))
+			return service.OpenAIErrorWrapperLocal(err, "check_request_sensitive_error", http.StatusBadRequest)
+		}
+	}
+
+	err = helper.ModelMappedHelper(c, relayInfo)
+	if err != nil {
+		return service.OpenAIErrorWrapperLocal(err, "model_mapped_error", http.StatusBadRequest)
+	}
+	req.Model = relayInfo.UpstreamModelName
+	if value, exists := c.Get("prompt_tokens"); exists {
+		promptTokens := value.(int)
+		relayInfo.SetPromptTokens(promptTokens)
+	} else {
+		promptTokens, err := getInputTokens(req, relayInfo)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "count_input_tokens_error", http.StatusBadRequest)
+		}
+		c.Set("prompt_tokens", promptTokens)
+	}
+
+	priceData, err := helper.ModelPriceHelper(c, relayInfo, relayInfo.PromptTokens, int(req.MaxOutputTokens))
+	if err != nil {
+		return service.OpenAIErrorWrapperLocal(err, "model_price_error", http.StatusInternalServerError)
+	}
+	// pre consume quota
+	preConsumedQuota, userQuota, openaiErr := preConsumeQuota(c, priceData.ShouldPreConsumedQuota, relayInfo)
+	if openaiErr != nil {
+		return openaiErr
+	}
+	defer func() {
+		if openaiErr != nil {
+			returnPreConsumedQuota(c, relayInfo, userQuota, preConsumedQuota)
+		}
+	}()
+	adaptor := GetAdaptor(relayInfo.ApiType)
+	if adaptor == nil {
+		return service.OpenAIErrorWrapperLocal(fmt.Errorf("invalid api type: %d", relayInfo.ApiType), "invalid_api_type", http.StatusBadRequest)
+	}
+	adaptor.Init(relayInfo)
+	var requestBody io.Reader
+	if model_setting.GetGlobalSettings().PassThroughRequestEnabled {
+		body, err := common.GetRequestBody(c)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "get_request_body_error", http.StatusInternalServerError)
+		}
+		requestBody = bytes.NewBuffer(body)
+	} else {
+		convertedRequest, err := adaptor.ConvertOpenAIResponsesRequest(c, relayInfo, *req)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "convert_request_error", http.StatusBadRequest)
+		}
+		jsonData, err := json.Marshal(convertedRequest)
+		if err != nil {
+			return service.OpenAIErrorWrapperLocal(err, "marshal_request_error", http.StatusInternalServerError)
+		}
+		// apply param override
+		if len(relayInfo.ParamOverride) > 0 {
+			reqMap := make(map[string]interface{})
+			err = json.Unmarshal(jsonData, &reqMap)
+			if err != nil {
+				return service.OpenAIErrorWrapperLocal(err, "param_override_unmarshal_failed", http.StatusInternalServerError)
+			}
+			for key, value := range relayInfo.ParamOverride {
+				reqMap[key] = value
+			}
+			jsonData, err = json.Marshal(reqMap)
+			if err != nil {
+				return service.OpenAIErrorWrapperLocal(err, "param_override_marshal_failed", http.StatusInternalServerError)
+			}
+		}
+
+		if common.DebugEnabled {
+			println("requestBody: ", string(jsonData))
+		}
+		requestBody = bytes.NewBuffer(jsonData)
+	}
+
+	var httpResp *http.Response
+	resp, err := adaptor.DoRequest(c, relayInfo, requestBody)
+	if err != nil {
+		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
+	}
+
+	statusCodeMappingStr := c.GetString("status_code_mapping")
+
+	if resp != nil {
+		httpResp = resp.(*http.Response)
+
+		if httpResp.StatusCode != http.StatusOK {
+			openaiErr = service.RelayErrorHandler(httpResp, false)
+			// reset status code 重置状态码
+			service.ResetStatusCode(openaiErr, statusCodeMappingStr)
+			return openaiErr
+		}
+	}
+
+	usage, openaiErr := adaptor.DoResponse(c, httpResp, relayInfo)
+	if openaiErr != nil {
+		// reset status code 重置状态码
+		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
+		return openaiErr
+	}
+
+	if strings.HasPrefix(relayInfo.OriginModelName, "gpt-4o-audio") {
+		service.PostAudioConsumeQuota(c, relayInfo, usage.(*dto.Usage), preConsumedQuota, userQuota, priceData, "")
+	} else {
+		postConsumeQuota(c, relayInfo, usage.(*dto.Usage), preConsumedQuota, userQuota, priceData, "")
+	}
+	return nil
+}

relay/relay-text.go

@@ -18,6 +18,7 @@ import (
 	"one-api/service"
 	"one-api/setting"
 	"one-api/setting/model_setting"
+	"one-api/setting/operation_setting"
 	"strings"
 	"time"
 
@@ -193,6 +194,7 @@ func TextHelper(c *gin.Context) (openaiErr *dto.OpenAIErrorWithStatusCode) {
 
 	var httpResp *http.Response
 	resp, err := adaptor.DoRequest(c, relayInfo, requestBody)
+
 	if err != nil {
 		return service.OpenAIErrorWrapper(err, "do_request_failed", http.StatusInternalServerError)
 	}
@@ -331,12 +333,14 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
 	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
 	promptTokens := usage.PromptTokens
 	cacheTokens := usage.PromptTokensDetails.CachedTokens
+	imageTokens := usage.PromptTokensDetails.ImageTokens
 	completionTokens := usage.CompletionTokens
 	modelName := relayInfo.OriginModelName
 
 	tokenName := ctx.GetString("token_name")
 	completionRatio := priceData.CompletionRatio
 	cacheRatio := priceData.CacheRatio
+	imageRatio := priceData.ImageRatio
 	modelRatio := priceData.ModelRatio
 	groupRatio := priceData.GroupRatio
 	modelPrice := priceData.ModelPrice
@@ -344,9 +348,11 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
 	// Convert values to decimal for precise calculation
 	dPromptTokens := decimal.NewFromInt(int64(promptTokens))
 	dCacheTokens := decimal.NewFromInt(int64(cacheTokens))
+	dImageTokens := decimal.NewFromInt(int64(imageTokens))
 	dCompletionTokens := decimal.NewFromInt(int64(completionTokens))
 	dCompletionRatio := decimal.NewFromFloat(completionRatio)
 	dCacheRatio := decimal.NewFromFloat(cacheRatio)
+	dImageRatio := decimal.NewFromFloat(imageRatio)
 	dModelRatio := decimal.NewFromFloat(modelRatio)
 	dGroupRatio := decimal.NewFromFloat(groupRatio)
 	dModelPrice := decimal.NewFromFloat(modelPrice)
@@ -354,11 +360,46 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
 
 	ratio := dModelRatio.Mul(dGroupRatio)
 
+	// openai web search 工具计费
+	var dWebSearchQuota decimal.Decimal
+	var webSearchPrice float64
+	if relayInfo.ResponsesUsageInfo != nil {
+		if webSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists && webSearchTool.CallCount > 0 {
+			// 计算 web search 调用的配额 (配额 = 价格 * 调用次数 / 1000 * 分组倍率)
+			webSearchPrice = operation_setting.GetWebSearchPricePerThousand(modelName, webSearchTool.SearchContextSize)
+			dWebSearchQuota = decimal.NewFromFloat(webSearchPrice).
+				Mul(decimal.NewFromInt(int64(webSearchTool.CallCount))).
+				Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+			extraContent += fmt.Sprintf("Web Search 调用 %d 次，上下文大小 %s，调用花费 $%s",
+				webSearchTool.CallCount, webSearchTool.SearchContextSize, dWebSearchQuota.String())
+		}
+	}
+	// file search tool 计费
+	var dFileSearchQuota decimal.Decimal
+	var fileSearchPrice float64
+	if relayInfo.ResponsesUsageInfo != nil {
+		if fileSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolFileSearch]; exists && fileSearchTool.CallCount > 0 {
+			fileSearchPrice = operation_setting.GetFileSearchPricePerThousand()
+			dFileSearchQuota = decimal.NewFromFloat(fileSearchPrice).
+				Mul(decimal.NewFromInt(int64(fileSearchTool.CallCount))).
+				Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+			extraContent += fmt.Sprintf("File Search 调用 %d 次，调用花费 $%s",
+				fileSearchTool.CallCount, dFileSearchQuota.String())
+		}
+	}
+
 	var quotaCalculateDecimal decimal.Decimal
 	if !priceData.UsePrice {
 		nonCachedTokens := dPromptTokens.Sub(dCacheTokens)
 		cachedTokensWithRatio := dCacheTokens.Mul(dCacheRatio)
+
 		promptQuota := nonCachedTokens.Add(cachedTokensWithRatio)
+		if imageTokens > 0 {
+			nonImageTokens := dPromptTokens.Sub(dImageTokens)
+			imageTokensWithRatio := dImageTokens.Mul(dImageRatio)
+			promptQuota = nonImageTokens.Add(imageTokensWithRatio)
+		}
+
 		completionQuota := dCompletionTokens.Mul(dCompletionRatio)
 
 		quotaCalculateDecimal = promptQuota.Add(completionQuota).Mul(ratio)
@@ -369,6 +410,9 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
 	} else {
 		quotaCalculateDecimal = dModelPrice.Mul(dQuotaPerUnit).Mul(dGroupRatio)
 	}
+	// 添加 responses tools call 调用的配额
+	quotaCalculateDecimal = quotaCalculateDecimal.Add(dWebSearchQuota)
+	quotaCalculateDecimal = quotaCalculateDecimal.Add(dFileSearchQuota)
 
 	quota := int(quotaCalculateDecimal.Round(0).IntPart())
 	totalTokens := promptTokens + completionTokens
@@ -414,6 +458,25 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
 		logContent += ", " + extraContent
 	}
 	other := service.GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, cacheTokens, cacheRatio, modelPrice)
+	if imageTokens != 0 {
+		other["image"] = true
+		other["image_ratio"] = imageRatio
+		other["image_output"] = imageTokens
+	}
+	if !dWebSearchQuota.IsZero() && relayInfo.ResponsesUsageInfo != nil {
+		if webSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists {
+			other["web_search"] = true
+			other["web_search_call_count"] = webSearchTool.CallCount
+			other["web_search_price"] = webSearchPrice
+		}
+	}
+	if !dFileSearchQuota.IsZero() && relayInfo.ResponsesUsageInfo != nil {
+		if fileSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolFileSearch]; exists {
+			other["file_search"] = true
+			other["file_search_call_count"] = fileSearchTool.CallCount
+			other["file_search_price"] = fileSearchPrice
+		}
+	}
 	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, promptTokens, completionTokens, logModel,
 		tokenName, quota, logContent, relayInfo.TokenId, userQuota, int(useTimeSeconds), relayInfo.IsStream, relayInfo.Group, other)
 }

router/relay-router.go

@@ -1,10 +1,11 @@
 package router
 
 import (
-	"github.com/gin-gonic/gin"
 	"one-api/controller"
 	"one-api/middleware"
 	"one-api/relay"
+
+	"github.com/gin-gonic/gin"
 )
 
 func SetRelayRouter(router *gin.Engine) {
@@ -40,13 +41,14 @@ func SetRelayRouter(router *gin.Engine) {
 		httpRouter.POST("/chat/completions", controller.Relay)
 		httpRouter.POST("/edits", controller.Relay)
 		httpRouter.POST("/images/generations", controller.Relay)
-		httpRouter.POST("/images/edits", controller.RelayNotImplemented)
+		httpRouter.POST("/images/edits", controller.Relay)
 		httpRouter.POST("/images/variations", controller.RelayNotImplemented)
 		httpRouter.POST("/embeddings", controller.Relay)
 		httpRouter.POST("/engines/:model/embeddings", controller.Relay)
 		httpRouter.POST("/audio/transcriptions", controller.Relay)
 		httpRouter.POST("/audio/translations", controller.Relay)
 		httpRouter.POST("/audio/speech", controller.Relay)
+		httpRouter.POST("/responses", controller.Relay)
 		httpRouter.GET("/files", controller.RelayNotImplemented)
 		httpRouter.POST("/files", controller.RelayNotImplemented)
 		httpRouter.DELETE("/files/:id", controller.RelayNotImplemented)

service/cf_worker.go

@@ -24,7 +24,7 @@ func DoWorkerRequest(req *WorkerRequest) (*http.Response, error) {
 	if !setting.EnableWorker() {
 		return nil, fmt.Errorf("worker not enabled")
 	}
-	if !strings.HasPrefix(req.URL, "https") {
+	if !setting.WorkerAllowHttpImageRequestEnabled && !strings.HasPrefix(req.URL, "https") {
 		return nil, fmt.Errorf("only support https url")
 	}
 

service/token_counter.go

@@ -400,6 +400,8 @@ func CountTokenMessages(info *relaycommon.RelayInfo, messages []dto.Message, mod
 					tokenNum += 100
 				} else if m.Type == dto.ContentTypeFile {
 					tokenNum += 5000
+				} else if m.Type == dto.ContentTypeVideoUrl {
+					tokenNum += 5000
 				} else {
 					tokenNum += getTokenNum(tokenEncoder, m.Text)
 				}

setting/operation_setting/model-ratio.go

@@ -51,26 +51,27 @@ var defaultModelRatio = map[string]float64{
 	"gpt-4o-realtime-preview-2024-12-17":      2.5,
 	"gpt-4o-mini-realtime-preview":            0.3,
 	"gpt-4o-mini-realtime-preview-2024-12-17": 0.3,
-	"o1":                         7.5,
-	"o1-2024-12-17":              7.5,
-	"o1-preview":                 7.5,
-	"o1-preview-2024-09-12":      7.5,
-	"o1-mini":                    0.55,
-	"o1-mini-2024-09-12":         0.55,
-	"o3-mini":                    0.55,
-	"o3-mini-2025-01-31":         0.55,
-	"o3-mini-high":               0.55,
-	"o3-mini-2025-01-31-high":    0.55,
-	"o3-mini-low":                0.55,
-	"o3-mini-2025-01-31-low":     0.55,
-	"o3-mini-medium":             0.55,
-	"o3-mini-2025-01-31-medium":  0.55,
-	"gpt-4o-mini":                0.075,
-	"gpt-4o-mini-2024-07-18":     0.075,
-	"gpt-4-turbo":                5, // $0.01 / 1K tokens
-	"gpt-4-turbo-2024-04-09":     5, // $0.01 / 1K tokens
-	"gpt-4.5-preview":            37.5,
-	"gpt-4.5-preview-2025-02-27": 37.5,
+	"gpt-image-1":                             2.5,
+	"o1":                                      7.5,
+	"o1-2024-12-17":                           7.5,
+	"o1-preview":                              7.5,
+	"o1-preview-2024-09-12":                   7.5,
+	"o1-mini":                                 0.55,
+	"o1-mini-2024-09-12":                      0.55,
+	"o3-mini":                                 0.55,
+	"o3-mini-2025-01-31":                      0.55,
+	"o3-mini-high":                            0.55,
+	"o3-mini-2025-01-31-high":                 0.55,
+	"o3-mini-low":                             0.55,
+	"o3-mini-2025-01-31-low":                  0.55,
+	"o3-mini-medium":                          0.55,
+	"o3-mini-2025-01-31-medium":               0.55,
+	"gpt-4o-mini":                             0.075,
+	"gpt-4o-mini-2024-07-18":                  0.075,
+	"gpt-4-turbo":                             5, // $0.01 / 1K tokens
+	"gpt-4-turbo-2024-04-09":                  5, // $0.01 / 1K tokens
+	"gpt-4.5-preview":                         37.5,
+	"gpt-4.5-preview-2025-02-27":              37.5,
 	//"gpt-3.5-turbo-0301":           0.75, //deprecated
 	"gpt-3.5-turbo":          0.25,
 	"gpt-3.5-turbo-0613":     0.75,
@@ -255,10 +256,11 @@ var defaultCompletionRatio = map[string]float64{
 	"gpt-4-gizmo-*":  2,
 	"gpt-4o-gizmo-*": 3,
 	"gpt-4-all":      2,
+	"gpt-image-1":    8,
 }
 
-// InitModelSettings initializes all model related settings maps
-func InitModelSettings() {
+// InitRatioSettings initializes all model related settings maps
+func InitRatioSettings() {
 	// Initialize modelPriceMap
 	modelPriceMapMutex.Lock()
 	modelPriceMap = defaultModelPrice
@@ -278,6 +280,12 @@ func InitModelSettings() {
 	cacheRatioMapMutex.Lock()
 	cacheRatioMap = defaultCacheRatio
 	cacheRatioMapMutex.Unlock()
+
+	// initialize imageRatioMap
+	imageRatioMapMutex.Lock()
+	imageRatioMap = defaultImageRatio
+	imageRatioMapMutex.Unlock()
+
 }
 
 func GetModelPriceMap() map[string]float64 {
@@ -509,18 +517,18 @@ func getHardcodedCompletionModelRatio(name string) (float64, bool) {
 
 func GetAudioRatio(name string) float64 {
 	if strings.Contains(name, "-realtime") {
-		if strings.HasSuffix(name, "gpt-4o-realtime-preview-2024-12-17") {
+		if strings.HasSuffix(name, "gpt-4o-realtime-preview") {
 			return 8
-		} else if strings.Contains(name, "mini") {
+		} else if strings.Contains(name, "gpt-4o-mini-realtime-preview") {
 			return 10 / 0.6
 		} else {
 			return 20
 		}
 	}
 	if strings.Contains(name, "-audio") {
-		if strings.HasSuffix(name, "gpt-4o-audio-preview-2024-12-17") {
-			return 16
-		} else if strings.Contains(name, "mini") {
+		if strings.HasPrefix(name, "gpt-4o-audio-preview") {
+			return 40 / 2.5
+		} else if strings.HasPrefix(name, "gpt-4o-mini-audio-preview") {
 			return 10 / 0.15
 		} else {
 			return 40
@@ -548,3 +556,36 @@ func ModelRatio2JSONString() string {
 	}
 	return string(jsonBytes)
 }
+
+var defaultImageRatio = map[string]float64{
+	"gpt-image-1": 2,
+}
+var imageRatioMap map[string]float64
+var imageRatioMapMutex sync.RWMutex
+
+func ImageRatio2JSONString() string {
+	imageRatioMapMutex.RLock()
+	defer imageRatioMapMutex.RUnlock()
+	jsonBytes, err := json.Marshal(imageRatioMap)
+	if err != nil {
+		common.SysError("error marshalling cache ratio: " + err.Error())
+	}
+	return string(jsonBytes)
+}
+
+func UpdateImageRatioByJSONString(jsonStr string) error {
+	imageRatioMapMutex.Lock()
+	defer imageRatioMapMutex.Unlock()
+	imageRatioMap = make(map[string]float64)
+	return json.Unmarshal([]byte(jsonStr), &imageRatioMap)
+}
+
+func GetImageRatio(name string) (float64, bool) {
+	imageRatioMapMutex.RLock()
+	defer imageRatioMapMutex.RUnlock()
+	ratio, ok := imageRatioMap[name]
+	if !ok {
+		return 1, false // Default to 1 if not found
+	}
+	return ratio, true
+}

setting/operation_setting/tools.go

@@ -0,0 +1,57 @@
+package operation_setting
+
+import "strings"
+
+const (
+	// Web search
+	WebSearchHighTierModelPriceLow    = 30.00
+	WebSearchHighTierModelPriceMedium = 35.00
+	WebSearchHighTierModelPriceHigh   = 50.00
+	WebSearchPriceLow                 = 25.00
+	WebSearchPriceMedium              = 27.50
+	WebSearchPriceHigh                = 30.00
+	// File search
+	FileSearchPrice = 2.5
+)
+
+func GetWebSearchPricePerThousand(modelName string, contextSize string) float64 {
+	// 确定模型类型
+	// https://platform.openai.com/docs/pricing Web search 价格按模型类型和 search context size 收费
+	// gpt-4.1, gpt-4o, or gpt-4o-search-preview 更贵，gpt-4.1-mini, gpt-4o-mini, gpt-4o-mini-search-preview 更便宜
+	isHighTierModel := (strings.HasPrefix(modelName, "gpt-4.1") || strings.HasPrefix(modelName, "gpt-4o")) &&
+		!strings.Contains(modelName, "mini")
+	// 确定 search context size 对应的价格
+	var priceWebSearchPerThousandCalls float64
+	switch contextSize {
+	case "low":
+		if isHighTierModel {
+			priceWebSearchPerThousandCalls = WebSearchHighTierModelPriceLow
+		} else {
+			priceWebSearchPerThousandCalls = WebSearchPriceLow
+		}
+	case "medium":
+		if isHighTierModel {
+			priceWebSearchPerThousandCalls = WebSearchHighTierModelPriceMedium
+		} else {
+			priceWebSearchPerThousandCalls = WebSearchPriceMedium
+		}
+	case "high":
+		if isHighTierModel {
+			priceWebSearchPerThousandCalls = WebSearchHighTierModelPriceHigh
+		} else {
+			priceWebSearchPerThousandCalls = WebSearchPriceHigh
+		}
+	default:
+		// search context size 默认为 medium
+		if isHighTierModel {
+			priceWebSearchPerThousandCalls = WebSearchHighTierModelPriceMedium
+		} else {
+			priceWebSearchPerThousandCalls = WebSearchPriceMedium
+		}
+	}
+	return priceWebSearchPerThousandCalls
+}
+
+func GetFileSearchPricePerThousand() float64 {
+	return FileSearchPrice
+}

setting/rate_limit.go

@@ -1,6 +1,64 @@
 package setting
 
+import (
+	"encoding/json"
+	"fmt"
+	"one-api/common"
+	"sync"
+)
+
 var ModelRequestRateLimitEnabled = false
 var ModelRequestRateLimitDurationMinutes = 1
 var ModelRequestRateLimitCount = 0
 var ModelRequestRateLimitSuccessCount = 1000
+var ModelRequestRateLimitGroup = map[string][2]int{}
+var ModelRequestRateLimitMutex sync.RWMutex
+
+func ModelRequestRateLimitGroup2JSONString() string {
+	ModelRequestRateLimitMutex.RLock()
+	defer ModelRequestRateLimitMutex.RUnlock()
+
+	jsonBytes, err := json.Marshal(ModelRequestRateLimitGroup)
+	if err != nil {
+		common.SysError("error marshalling model ratio: " + err.Error())
+	}
+	return string(jsonBytes)
+}
+
+func UpdateModelRequestRateLimitGroupByJSONString(jsonStr string) error {
+	ModelRequestRateLimitMutex.RLock()
+	defer ModelRequestRateLimitMutex.RUnlock()
+
+	ModelRequestRateLimitGroup = make(map[string][2]int)
+	return json.Unmarshal([]byte(jsonStr), &ModelRequestRateLimitGroup)
+}
+
+func GetGroupRateLimit(group string) (totalCount, successCount int, found bool) {
+	ModelRequestRateLimitMutex.RLock()
+	defer ModelRequestRateLimitMutex.RUnlock()
+
+	if ModelRequestRateLimitGroup == nil {
+		return 0, 0, false
+	}
+
+	limits, found := ModelRequestRateLimitGroup[group]
+	if !found {
+		return 0, 0, false
+	}
+	return limits[0], limits[1], true
+}
+
+func CheckModelRequestRateLimitGroup(jsonStr string) error {
+	checkModelRequestRateLimitGroup := make(map[string][2]int)
+	err := json.Unmarshal([]byte(jsonStr), &checkModelRequestRateLimitGroup)
+	if err != nil {
+		return err
+	}
+	for group, limits := range checkModelRequestRateLimitGroup {
+		if limits[0] < 0 || limits[1] < 1 {
+			return fmt.Errorf("group %s has negative rate limit values: [%d, %d]", group, limits[0], limits[1])
+		}
+	}
+
+	return nil
+}

setting/system_setting.go

@@ -3,6 +3,7 @@ package setting
 var ServerAddress = "http://localhost:3000"
 var WorkerUrl = ""
 var WorkerValidKey = ""
+var WorkerAllowHttpImageRequestEnabled = false
 
 func EnableWorker() bool {
 	return WorkerUrl != ""

web/src/components/ChannelsTable.js

@@ -1276,7 +1276,7 @@ const ChannelsTable = () => {
             <Form.Input
               field='search_keyword'
               label={t('搜索渠道关键词')}
-              placeholder={t('搜索渠道的 ID，名称和密钥 ...')}
+              placeholder={t('搜索渠道的 ID，名称，密钥和API地址 ...')}
               value={searchKeyword}
               loading={searching}
               onChange={(v) => {

web/src/components/LogsTable.js

@@ -103,9 +103,15 @@ const LogsTable = () => {
             {t('系统')}
           </Tag>
         );
+      case 5:
+        return (
+          <Tag color='red' size='large'>
+            {t('错误')}
+          </Tag>
+        );
       default:
         return (
-          <Tag color='black' size='large'>
+          <Tag color='grey' size='large'>
             {t('未知')}
           </Tag>
         );
@@ -373,7 +379,7 @@ const LogsTable = () => {
       className: isAdmin() ? 'tableShow' : 'tableHiddle',
       render: (text, record, index) => {
         return isAdminUser ? (
-          record.type === 0 || record.type === 2 ? (
+          record.type === 0 || record.type === 2 || record.type === 5 ? (
             <div>
               {
                 <Tooltip content={record.channel_name || '[未知]'}>
@@ -426,7 +432,7 @@ const LogsTable = () => {
       title: t('令牌'),
       dataIndex: 'token_name',
       render: (text, record, index) => {
-        return record.type === 0 || record.type === 2 ? (
+        return record.type === 0 || record.type === 2 || record.type === 5 ? (
           <div>
             <Tag
               color='grey'
@@ -450,7 +456,7 @@ const LogsTable = () => {
       title: t('分组'),
       dataIndex: 'group',
       render: (text, record, index) => {
-        if (record.type === 0 || record.type === 2) {
+        if (record.type === 0 || record.type === 2 || record.type === 5) {
           if (record.group) {
             return <>{renderGroup(record.group)}</>;
           } else {
@@ -490,7 +496,7 @@ const LogsTable = () => {
       title: t('模型'),
       dataIndex: 'model_name',
       render: (text, record, index) => {
-        return record.type === 0 || record.type === 2 ? (
+        return record.type === 0 || record.type === 2 || record.type === 5 ? (
           <>{renderModelName(record)}</>
         ) : (
           <></>
@@ -530,7 +536,7 @@ const LogsTable = () => {
       title: t('提示'),
       dataIndex: 'prompt_tokens',
       render: (text, record, index) => {
-        return record.type === 0 || record.type === 2 ? (
+        return record.type === 0 || record.type === 2 || record.type === 5 ? (
           <>{<span> {text} </span>}</>
         ) : (
           <></>
@@ -543,7 +549,7 @@ const LogsTable = () => {
       dataIndex: 'completion_tokens',
       render: (text, record, index) => {
         return parseInt(text) > 0 &&
-          (record.type === 0 || record.type === 2) ? (
+          (record.type === 0 || record.type === 2 || record.type === 5) ? (
           <>{<span> {text} </span>}</>
         ) : (
           <></>
@@ -555,7 +561,7 @@ const LogsTable = () => {
       title: t('花费'),
       dataIndex: 'quota',
       render: (text, record, index) => {
-        return record.type === 0 || record.type === 2 ? (
+        return record.type === 0 || record.type === 2 || record.type === 5 ? (
           <>{renderQuota(text, 6)}</>
         ) : (
           <></>
@@ -612,7 +618,6 @@ const LogsTable = () => {
             </Paragraph>
           );
         }
-
         let content = other?.claude
           ? renderClaudeModelPriceSimple(
               other.model_ratio,
@@ -920,7 +925,6 @@ const LogsTable = () => {
                 other.completion_ratio,
                 other.model_price,
                 other.group_ratio,
-                other.user_group_ratio,
                 other.cache_ratio || 1.0,
                 other.cache_creation_ratio || 1.0,
               )
@@ -929,7 +933,14 @@ const LogsTable = () => {
                 other.completion_ratio,
                 other.model_price,
                 other.group_ratio,
-                other.user_group_ratio,
+                other?.user_group_ratio,
+                false,
+                1.0,
+                undefined,
+                other.web_search || false,
+                other.web_search_call_count || 0,
+                other.file_search || false,
+                other.file_search_call_count || 0,
               ),
         });
       }
@@ -987,6 +998,15 @@ const LogsTable = () => {
             other?.group_ratio,
             other?.cache_tokens || 0,
             other?.cache_ratio || 1.0,
+            other?.image || false,
+            other?.image_ratio || 0,
+            other?.image_output || 0,
+            other?.web_search || false,
+            other?.web_search_call_count || 0,
+            other?.web_search_price || 0,
+            other?.file_search || false,
+            other?.file_search_call_count || 0,
+            other?.file_search_price || 0,
           );
         }
         expandDataLocal.push({
@@ -1246,6 +1266,7 @@ const LogsTable = () => {
             <Select.Option value='2'>{t('消费')}</Select.Option>
             <Select.Option value='3'>{t('管理')}</Select.Option>
             <Select.Option value='4'>{t('系统')}</Select.Option>
+            <Select.Option value='5'>{t('错误')}</Select.Option>
           </Select>
           <Button
             theme='light'

web/src/components/ModelPricing.js

@@ -81,7 +81,7 @@ const ModelPricing = () => {
   }
 
   function renderAvailable(available) {
-    return (
+    return available ? (
       <Popover
         content={
           <div style={{ padding: 8 }}>{t('您的分组可以使用该模型')}</div>
@@ -98,7 +98,7 @@ const ModelPricing = () => {
       >
         <IconVerify style={{ color: 'green' }} size='large' />
       </Popover>
-    );
+    ) : null;
   }
 
   const columns = [
@@ -109,7 +109,12 @@ const ModelPricing = () => {
         // if record.enable_groups contains selectedGroup, then available is true
         return renderAvailable(record.enable_groups.includes(selectedGroup));
       },
-      sorter: (a, b) => a.available - b.available,
+      sorter: (a, b) => {
+        const aAvailable = a.enable_groups.includes(selectedGroup);
+        const bAvailable = b.enable_groups.includes(selectedGroup);
+        return Number(aAvailable) - Number(bAvailable);
+      },
+      defaultSortOrder: 'descend',
     },
     {
       title: t('模型名称'),

web/src/components/PersonalSetting.js

@@ -57,6 +57,7 @@ const PersonalSetting = () => {
     email_verification_code: '',
     email: '',
     self_account_deletion_confirmation: '',
+    original_password: '',
     set_new_password: '',
     set_new_password_confirmation: '',
   });
@@ -239,11 +240,24 @@ const PersonalSetting = () => {
   };
 
   const changePassword = async () => {
+    if (inputs.original_password === '') {
+      showError(t('请输入原密码！'));
+      return;
+    }
+    if (inputs.set_new_password === '') {
+      showError(t('请输入新密码！'));
+      return;
+    }
+    if (inputs.original_password === inputs.set_new_password) {
+      showError(t('新密码需要和原密码不一致！'));
+      return;
+    }
     if (inputs.set_new_password !== inputs.set_new_password_confirmation) {
       showError(t('两次输入的密码不一致！'));
       return;
     }
     const res = await API.put(`/api/user/self`, {
+      original_password: inputs.original_password,
       password: inputs.set_new_password,
     });
     const { success, message } = res.data;
@@ -816,8 +830,8 @@ const PersonalSetting = () => {
               </div>
             </Card>
             <Card style={{ marginTop: 10 }}>
-              <Tabs type="line" defaultActiveKey="notification">
-                <TabPane tab={t('通知设置')} itemKey="notification">
+              <Tabs type='line' defaultActiveKey='notification'>
+                <TabPane tab={t('通知设置')} itemKey='notification'>
                   <div style={{ marginTop: 20 }}>
                     <Typography.Text strong>{t('通知方式')}</Typography.Text>
                     <div style={{ marginTop: 10 }}>
@@ -993,23 +1007,36 @@ const PersonalSetting = () => {
                     </Typography.Text>
                   </div>
                 </TabPane>
-                <TabPane tab={t('价格设置')} itemKey="price">
+                <TabPane tab={t('价格设置')} itemKey='price'>
                   <div style={{ marginTop: 20 }}>
-                    <Typography.Text strong>{t('接受未设置价格模型')}</Typography.Text>
+                    <Typography.Text strong>
+                      {t('接受未设置价格模型')}
+                    </Typography.Text>
                     <div style={{ marginTop: 10 }}>
                       <Checkbox
-                        checked={notificationSettings.acceptUnsetModelRatioModel}
-                        onChange={e => handleNotificationSettingChange('acceptUnsetModelRatioModel', e.target.checked)}
+                        checked={
+                          notificationSettings.acceptUnsetModelRatioModel
+                        }
+                        onChange={(e) =>
+                          handleNotificationSettingChange(
+                            'acceptUnsetModelRatioModel',
+                            e.target.checked,
+                          )
+                        }
                       >
                         {t('接受未设置价格模型')}
                       </Checkbox>
-                      <Typography.Text type="secondary" style={{ marginTop: 8, display: 'block' }}>
-                        {t('当模型没有设置价格时仍接受调用，仅当您信任该网站时使用，可能会产生高额费用')}
+                      <Typography.Text
+                        type='secondary'
+                        style={{ marginTop: 8, display: 'block' }}
+                      >
+                        {t(
+                          '当模型没有设置价格时仍接受调用，仅当您信任该网站时使用，可能会产生高额费用',
+                        )}
                       </Typography.Text>
                     </div>
                   </div>
                 </TabPane>
-
               </Tabs>
               <div style={{ marginTop: 20 }}>
                 <Button type='primary' onClick={saveNotificationSettings}>
@@ -1118,6 +1145,16 @@ const PersonalSetting = () => {
             >
               <div style={{ marginTop: 20 }}>
                 <Input
+                  name='original_password'
+                  placeholder={t('原密码')}
+                  type='password'
+                  value={inputs.original_password}
+                  onChange={(value) =>
+                    handleInputChange('original_password', value)
+                  }
+                />
+                <Input
+                  style={{ marginTop: 20 }}
                   name='set_new_password'
                   placeholder={t('新密码')}
                   value={inputs.set_new_password}

web/src/components/RateLimitSetting.js

@@ -13,6 +13,7 @@ const RateLimitSetting = () => {
     ModelRequestRateLimitCount: 0,
     ModelRequestRateLimitSuccessCount: 1000,
     ModelRequestRateLimitDurationMinutes: 1,
+    ModelRequestRateLimitGroup: '',
   });
 
   let [loading, setLoading] = useState(false);
@@ -23,10 +24,14 @@ const RateLimitSetting = () => {
     if (success) {
       let newInputs = {};
       data.forEach((item) => {
-        if (item.key.endsWith('Enabled')) {
-          newInputs[item.key] = item.value === 'true' ? true : false;
-        } else {
-          newInputs[item.key] = item.value;
+      if (item.key === 'ModelRequestRateLimitGroup') {
+        item.value = JSON.stringify(JSON.parse(item.value), null, 2);
+      }
+
+      if (item.key.endsWith('Enabled')) {
+        newInputs[item.key] = item.value === 'true' ? true : false;
+      } else {
+        newInputs[item.key] = item.value;
         }
       });
 

web/src/components/SystemSetting.js

@@ -19,6 +19,7 @@ import {
   verifyJSON,
 } from '../helpers/utils';
 import { API } from '../helpers/api';
+import axios from 'axios';
 
 const SystemSetting = () => {
   let [inputs, setInputs] = useState({
@@ -44,6 +45,7 @@ const SystemSetting = () => {
     ServerAddress: '',
     WorkerUrl: '',
     WorkerValidKey: '',
+    WorkerAllowHttpImageRequestEnabled: '',
     EpayId: '',
     EpayKey: '',
     Price: 7.3,
@@ -110,6 +112,7 @@ const SystemSetting = () => {
           case 'SMTPSSLEnabled':
           case 'LinuxDOOAuthEnabled':
           case 'oidc.enabled':
+          case 'WorkerAllowHttpImageRequestEnabled':
             item.value = item.value === 'true';
             break;
           case 'Price':
@@ -203,10 +206,17 @@ const SystemSetting = () => {
 
   const submitWorker = async () => {
     let WorkerUrl = removeTrailingSlash(inputs.WorkerUrl);
-    await updateOptions([
+    const options = [
       { key: 'WorkerUrl', value: WorkerUrl },
-      { key: 'WorkerValidKey', value: inputs.WorkerValidKey },
-    ]);
+      {
+        key: 'WorkerAllowHttpImageRequestEnabled',
+        value: inputs.WorkerAllowHttpImageRequestEnabled ? 'true' : 'false',
+      },
+    ];
+    if (inputs.WorkerValidKey !== '' || WorkerUrl === '') {
+      options.push({ key: 'WorkerValidKey', value: inputs.WorkerValidKey });
+    }
+    await updateOptions(options);
   };
 
   const submitPayAddress = async () => {
@@ -298,7 +308,8 @@ const SystemSetting = () => {
       const domain = emailToAdd.trim();
 
       // 验证域名格式
-      const domainRegex = /^([a-zA-Z0-9]([a-zA-Z0-9\-]{0,61}[a-zA-Z0-9])?\.)+[a-zA-Z]{2,}$/;
+      const domainRegex =
+        /^([a-zA-Z0-9]([a-zA-Z0-9\-]{0,61}[a-zA-Z0-9])?\.)+[a-zA-Z]{2,}$/;
       if (!domainRegex.test(domain)) {
         showError('邮箱域名格式不正确，请输入有效的域名，如 gmail.com');
         return;
@@ -371,7 +382,7 @@ const SystemSetting = () => {
   };
 
   const submitOIDCSettings = async () => {
-    if (inputs['oidc.well_known'] !== '') {
+    if (inputs['oidc.well_known'] && inputs['oidc.well_known'] !== '') {
       if (
         !inputs['oidc.well_known'].startsWith('http://') &&
         !inputs['oidc.well_known'].startsWith('https://')
@@ -380,7 +391,7 @@ const SystemSetting = () => {
         return;
       }
       try {
-        const res = await API.get(inputs['oidc.well_known']);
+        const res = await axios.create().get(inputs['oidc.well_known']);
         inputs['oidc.authorization_endpoint'] =
           res.data['authorization_endpoint'];
         inputs['oidc.token_endpoint'] = res.data['token_endpoint'];
@@ -573,6 +584,12 @@ const SystemSetting = () => {
                       />
                     </Col>
                   </Row>
+                  <Form.Checkbox
+                    field='WorkerAllowHttpImageRequestEnabled'
+                    noLabel
+                  >
+                    允许 HTTP 协议图片请求（适用于自部署代理）
+                  </Form.Checkbox>
                   <Button onClick={submitWorker}>更新Worker设置</Button>
                 </Form.Section>
               </Card>
@@ -795,7 +812,13 @@ const SystemSetting = () => {
                     onChange={(value) => setEmailToAdd(value)}
                     style={{ marginTop: 16 }}
                     suffix={
-                      <Button theme="solid" type="primary" onClick={handleAddEmail}>添加</Button>
+                      <Button
+                        theme='solid'
+                        type='primary'
+                        onClick={handleAddEmail}
+                      >
+                        添加
+                      </Button>
                     }
                     onEnterPress={handleAddEmail}
                   />

web/src/helpers/render.js

@@ -314,6 +314,15 @@ export function renderModelPrice(
   groupRatio,
   cacheTokens = 0,
   cacheRatio = 1.0,
+  image = false,
+  imageRatio = 1.0,
+  imageOutputTokens = 0,
+  webSearch = false,
+  webSearchCallCount = 0,
+  webSearchPrice = 0,
+  fileSearch = false,
+  fileSearchCallCount = 0,
+  fileSearchPrice = 0,
 ) {
   if (modelPrice !== -1) {
     return i18next.t(
@@ -331,26 +340,34 @@ export function renderModelPrice(
     let inputRatioPrice = modelRatio * 2.0;
     let completionRatioPrice = modelRatio * 2.0 * completionRatio;
     let cacheRatioPrice = modelRatio * 2.0 * cacheRatio;
+    let imageRatioPrice = modelRatio * 2.0 * imageRatio;
 
     // Calculate effective input tokens (non-cached + cached with ratio applied)
-    const effectiveInputTokens =
+    let effectiveInputTokens =
       inputTokens - cacheTokens + cacheTokens * cacheRatio;
+    // Handle image tokens if present
+    if (image && imageOutputTokens > 0) {
+      effectiveInputTokens =
+        inputTokens - imageOutputTokens + imageOutputTokens * imageRatio;
+    }
 
     let price =
       (effectiveInputTokens / 1000000) * inputRatioPrice * groupRatio +
-      (completionTokens / 1000000) * completionRatioPrice * groupRatio;
+      (completionTokens / 1000000) * completionRatioPrice * groupRatio +
+      (webSearchCallCount / 1000) * webSearchPrice * groupRatio +
+      (fileSearchCallCount / 1000) * fileSearchPrice * groupRatio;
 
     return (
       <>
         <article>
           <p>
-            {i18next.t('提示价格：${{price}} / 1M tokens', {
+            {i18next.t('输入价格：${{price}} / 1M tokens', {
               price: inputRatioPrice,
             })}
           </p>
           <p>
             {i18next.t(
-              '补全价格：${{price}} * {{completionRatio}} = ${{total}} / 1M tokens (补全倍率: {{completionRatio}})',
+              '输出价格：${{price}} * {{completionRatio}} = ${{total}} / 1M tokens (补全倍率: {{completionRatio}})',
               {
                 price: inputRatioPrice,
                 total: completionRatioPrice,
@@ -370,11 +387,38 @@ export function renderModelPrice(
               )}
             </p>
           )}
+          {image && imageOutputTokens > 0 && (
+            <p>
+              {i18next.t(
+                '图片输入价格：${{price}} * {{ratio}} = ${{total}} / 1M tokens (图片倍率: {{imageRatio}})',
+                {
+                  price: imageRatioPrice,
+                  ratio: groupRatio,
+                  total: imageRatioPrice * groupRatio,
+                  imageRatio: imageRatio,
+                },
+              )}
+            </p>
+          )}
+          {webSearch && webSearchCallCount > 0 && (
+            <p>
+              {i18next.t('Web搜索价格：${{price}} / 1K 次', {
+                price: webSearchPrice,
+              })}
+            </p>
+          )}
+          {fileSearch && fileSearchCallCount > 0 && (
+            <p>
+              {i18next.t('文件搜索价格：${{price}} / 1K 次', {
+                price: fileSearchPrice,
+              })}
+            </p>
+          )}
           <p></p>
           <p>
-            {cacheTokens > 0
+            {cacheTokens > 0 && !image && !webSearch && !fileSearch
               ? i18next.t(
-                  '提示 {{nonCacheInput}} tokens / 1M tokens * ${{price}} + 缓存 {{cacheInput}} tokens / 1M tokens * ${{cachePrice}} + 补全 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}',
+                  '输入 {{nonCacheInput}} tokens / 1M tokens * ${{price}} + 缓存 {{cacheInput}} tokens / 1M tokens * ${{cachePrice}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}',
                   {
                     nonCacheInput: inputTokens - cacheTokens,
                     cacheInput: cacheTokens,
@@ -386,17 +430,82 @@ export function renderModelPrice(
                     total: price.toFixed(6),
                   },
                 )
-              : i18next.t(
-                  '提示 {{input}} tokens / 1M tokens * ${{price}} + 补全 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}',
-                  {
-                    input: inputTokens,
-                    price: inputRatioPrice,
-                    completion: completionTokens,
-                    compPrice: completionRatioPrice,
-                    ratio: groupRatio,
-                    total: price.toFixed(6),
-                  },
-                )}
+              : image && imageOutputTokens > 0 && !webSearch && !fileSearch
+                ? i18next.t(
+                    '输入 {{nonImageInput}} tokens + 图片输入 {{imageInput}} tokens * {{imageRatio}} / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}',
+                    {
+                      nonImageInput: inputTokens - imageOutputTokens,
+                      imageInput: imageOutputTokens,
+                      imageRatio: imageRatio,
+                      price: inputRatioPrice,
+                      completion: completionTokens,
+                      compPrice: completionRatioPrice,
+                      ratio: groupRatio,
+                      total: price.toFixed(6),
+                    },
+                  )
+                : webSearch && webSearchCallCount > 0 && !image && !fileSearch
+                  ? i18next.t(
+                      '输入 {{input}} tokens / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} + Web搜索 {{webSearchCallCount}}次 / 1K 次 * ${{webSearchPrice}} * {{ratio}} = ${{total}}',
+                      {
+                        input: inputTokens,
+                        price: inputRatioPrice,
+                        completion: completionTokens,
+                        compPrice: completionRatioPrice,
+                        ratio: groupRatio,
+                        webSearchCallCount,
+                        webSearchPrice,
+                        total: price.toFixed(6),
+                      },
+                    )
+                  : fileSearch &&
+                      fileSearchCallCount > 0 &&
+                      !image &&
+                      !webSearch
+                    ? i18next.t(
+                        '输入 {{input}} tokens / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} + 文件搜索 {{fileSearchCallCount}}次 / 1K 次 * ${{fileSearchPrice}} * {{ratio}}= ${{total}}',
+                        {
+                          input: inputTokens,
+                          price: inputRatioPrice,
+                          completion: completionTokens,
+                          compPrice: completionRatioPrice,
+                          ratio: groupRatio,
+                          fileSearchCallCount,
+                          fileSearchPrice,
+                          total: price.toFixed(6),
+                        },
+                      )
+                    : webSearch &&
+                        webSearchCallCount > 0 &&
+                        fileSearch &&
+                        fileSearchCallCount > 0 &&
+                        !image
+                      ? i18next.t(
+                          '输入 {{input}} tokens / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} + Web搜索 {{webSearchCallCount}}次 / 1K 次 * ${{webSearchPrice}} * {{ratio}}+ 文件搜索 {{fileSearchCallCount}}次 / 1K 次 * ${{fileSearchPrice}} * {{ratio}}= ${{total}}',
+                          {
+                            input: inputTokens,
+                            price: inputRatioPrice,
+                            completion: completionTokens,
+                            compPrice: completionRatioPrice,
+                            ratio: groupRatio,
+                            webSearchCallCount,
+                            webSearchPrice,
+                            fileSearchCallCount,
+                            fileSearchPrice,
+                            total: price.toFixed(6),
+                          },
+                        )
+                      : i18next.t(
+                          '输入 {{input}} tokens / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}',
+                          {
+                            input: inputTokens,
+                            price: inputRatioPrice,
+                            completion: completionTokens,
+                            compPrice: completionRatioPrice,
+                            ratio: groupRatio,
+                            total: price.toFixed(6),
+                          },
+                        )}
           </p>
           <p>{i18next.t('仅供参考，以实际扣费为准')}</p>
         </article>
@@ -405,12 +514,76 @@ export function renderModelPrice(
   }
 }
 
+export function renderLogContent(
+  modelRatio,
+  completionRatio,
+  modelPrice = -1,
+  groupRatio,
+  user_group_ratio,
+  image = false,
+  imageRatio = 1.0,
+  useUserGroupRatio = undefined,
+  webSearch = false,
+  webSearchCallCount = 0,
+  fileSearch = false,
+  fileSearchCallCount = 0,
+) {
+  const ratioLabel = useUserGroupRatio
+    ? i18next.t('专属倍率')
+    : i18next.t('分组倍率');
+  const ratio = useUserGroupRatio ? user_group_ratio : groupRatio;
+
+  if (modelPrice !== -1) {
+    return i18next.t('模型价格 ${{price}}，{{ratioType}} {{ratio}}', {
+      price: modelPrice,
+      ratioType: ratioLabel,
+      ratio,
+    });
+  } else {
+    if (image) {
+      return i18next.t(
+        '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，图片输入倍率 {{imageRatio}}，{{ratioType}} {{ratio}}',
+        {
+          modelRatio: modelRatio,
+          completionRatio: completionRatio,
+          imageRatio: imageRatio,
+          ratioType: ratioLabel,
+          ratio,
+        },
+      );
+    } else if (webSearch) {
+      return i18next.t(
+        '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，{{ratioType}} {{ratio}}，Web 搜索调用 {{webSearchCallCount}} 次',
+        {
+          modelRatio: modelRatio,
+          completionRatio: completionRatio,
+          ratioType: ratioLabel,
+          ratio,
+          webSearchCallCount,
+        },
+      );
+    } else {
+      return i18next.t(
+        '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，{{ratioType}} {{ratio}}',
+        {
+          modelRatio: modelRatio,
+          completionRatio: completionRatio,
+          ratioType: ratioLabel,
+          ratio,
+        },
+      );
+    }
+  }
+}
+
 export function renderModelPriceSimple(
   modelRatio,
   modelPrice = -1,
   groupRatio,
   cacheTokens = 0,
   cacheRatio = 1.0,
+  image = false,
+  imageRatio = 1.0,
 ) {
   if (modelPrice !== -1) {
     return i18next.t('价格：${{price}} * 分组：{{ratio}}', {
@@ -418,7 +591,28 @@ export function renderModelPriceSimple(
       ratio: groupRatio,
     });
   } else {
-    if (cacheTokens !== 0) {
+    if (image && cacheTokens !== 0) {
+      return i18next.t(
+        '模型: {{ratio}} * {{ratioType}}: {{groupRatio}} * 缓存倍率: {{cacheRatio}} * 图片输入倍率: {{imageRatio}}',
+        {
+          ratio: modelRatio,
+          ratioType: ratioLabel,
+          groupRatio: groupRatio,
+          cacheRatio: cacheRatio,
+          imageRatio: imageRatio,
+        },
+      );
+    } else if (image) {
+      return i18next.t(
+        '模型: {{ratio}} * {{ratioType}}: {{groupRatio}} * 图片输入倍率: {{imageRatio}}',
+        {
+          ratio: modelRatio,
+          ratioType: ratioLabel,
+          groupRatio: groupRatio,
+          imageRatio: imageRatio,
+        },
+      );
+    } else if (cacheTokens !== 0) {
       return i18next.t(
         '模型: {{ratio}} * 分组: {{groupRatio}} * 缓存: {{cacheRatio}}',
         {
@@ -882,7 +1076,7 @@ export function renderClaudeLogContent(
     });
   } else {
     return i18next.t(
-      '模型倍率 {{modelRatio}}，补全倍率 {{completionRatio}}，缓存倍率 {{cacheRatio}}，缓存创建倍率 {{cacheCreationRatio}}，{{ratioType}} {{ratio}}',
+      '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，缓存倍率 {{cacheRatio}}，缓存创建倍率 {{cacheCreationRatio}}，{{ratioType}} {{ratio}}',
       {
         modelRatio: modelRatio,
         completionRatio: completionRatio,
@@ -933,30 +1127,3 @@ export function renderClaudeModelPriceSimple(
     }
   }
 }
-
-export function renderLogContent(
-  modelRatio,
-  completionRatio,
-  modelPrice = -1,
-  groupRatio,
-) {
-  const ratioLabel = false ? i18next.t('专属倍率') : i18next.t('分组倍率');
-
-  if (modelPrice !== -1) {
-    return i18next.t('模型价格 ${{price}}，{{ratioType}} {{ratio}}', {
-      price: modelPrice,
-      ratioType: ratioLabel,
-      ratio: groupRatio,
-    });
-  } else {
-    return i18next.t(
-      '模型倍率 {{modelRatio}}，补全倍率 {{completionRatio}}，{{ratioType}} {{ratio}}',
-      {
-        modelRatio: modelRatio,
-        completionRatio: completionRatio,
-        ratioType: ratioLabel,
-        ratio: groupRatio,
-      },
-    );
-  }
-}

web/src/helpers/utils.js

@@ -143,6 +143,7 @@ export function openPage(url) {
 }
 
 export function removeTrailingSlash(url) {
+  if (!url) return '';
   if (url.endsWith('/')) {
     return url.slice(0, -1);
   } else {

web/src/i18n/locales/en.json

@@ -124,7 +124,7 @@
   "已成功开始测试所有已启用通道，请刷新页面查看结果。": "Successfully started testing all enabled channels. Please refresh page to view results.",
   "通道 ${name} 余额更新成功！": "Channel ${name} quota updated successfully!",
   "已更新完毕所有已启用通道余额！": "Updated quota for all enabled channels!",
-  "搜索渠道的 ID，名称和密钥 ...": "Search channel ID, name and key...",
+  "搜索渠道的 ID，名称，密钥和API地址 ...": "Search channel ID, name, key and Base URL...",
   "名称": "Name",
   "分组": "Group",
   "类型": "Type",
@@ -492,7 +492,8 @@
   "请输入默认 API 版本，例如：2023-03-15-preview，该配置可以被实际的请求查询参数所覆盖": "Please enter the default API version, for example: 2023-03-15-preview, this configuration can be overridden by the actual request query parameters",
   "默认": "default",
   "图片演示": "Image demo",
-  "注意，系统请求的时模型名称中的点会被剔除，例如：gpt-4.5-preview会请求为gpt-45-preview，所以部署的模型名称需要去掉点": "Note that the dot in the model name requested by the system will be removed, for example: gpt-4.5-preview will be requested as gpt-45-preview, so the deployed model name needs to remove the dot",
+  "注意，系统请求的时模型名称中的点会被剔除，例如：gpt-4.1会请求为gpt-41，所以在Azure部署的时候，部署模型名称需要手动改为gpt-41": "Note that the dot in the model name requested by the system will be removed, for example: gpt-4.1 will be requested as gpt-41, so when deploying on Azure, the deployment model name needs to be manually changed to gpt-41",
+  "2025年5月10日后添加的渠道，不需要再在部署的时候移除模型名称中的\".\"": "After May 10, 2025, channels added do not need to remove the dot in the model name during deployment",
   "模型映射必须是合法的 JSON 格式！": "Model mapping must be in valid JSON format!",
   "取消无限额度": "Cancel unlimited quota",
   "取消": "Cancel",
@@ -679,7 +680,10 @@
   "当前分组可用": "Available in current group",
   "当前分组不可用": "The current group is unavailable",
   "提示：": "input:",
+  "输入：": "input:",
   "补全：": "output:",
+  "输出：": "output:",
+  "图片输出：": "Image output:",
   "模型价格：": "Model price:",
   "模型：": "Model:",
   "分组：": "Grouping:",
@@ -1054,14 +1058,16 @@
   "等级": "grade",
   "钉钉": "DingTalk",
   "模型价格：${{price}} * 分组倍率：{{ratio}} = ${{total}}": "Model price: ${{price}} * Group ratio: {{ratio}} = ${{total}}",
-  "提示：${{price}} * {{ratio}} = ${{total}} / 1M tokens": "Prompt: ${{price}} * {{ratio}} = ${{total}} / 1M tokens",
-  "补全：${{price}} * {{ratio}} = ${{total}} / 1M tokens": "Completion: ${{price}} * {{ratio}} = ${{total}} / 1M tokens",
-  "音频提示：${{price}} * {{ratio}} * {{audioRatio}} = ${{total}} / 1M tokens": "Audio prompt: ${{price}} * {{ratio}} * {{audioRatio}} = ${{total}} / 1M tokens",
+  "输入：${{price}} * {{ratio}} = ${{total}} / 1M tokens": "Prompt: ${{price}} * {{ratio}} = ${{total}} / 1M tokens",
+  "输出：${{price}} * {{ratio}} = ${{total}} / 1M tokens": "Completion: ${{price}} * {{ratio}} = ${{total}} / 1M tokens",
+  "图片输入：${{price}} * {{ratio}} = ${{total}} / 1M tokens (图片倍率: {{imageRatio}})": "Image input: ${{price}} * {{ratio}} = ${{total}} / 1M tokens (Image ratio: {{imageRatio}})",
+  "音频输入：${{price}} * {{ratio}} * {{audioRatio}} = ${{total}} / 1M tokens": "Audio prompt: ${{price}} * {{ratio}} * {{audioRatio}} = ${{total}} / 1M tokens",
   "音频提示 {{input}} tokens / 1M tokens * ${{price}} * {{audioRatio}} + 音频补全 {{completion}} tokens / 1M tokens * ${{price}} * {{audioRatio}} * {{audioCompRatio}}": "Audio prompt {{input}} tokens / 1M tokens * ${{price}} * {{audioRatio}} + Audio completion {{completion}} tokens / 1M tokens * ${{price}} * {{audioRatio}} * {{audioCompRatio}}",
-  "音频补全：${{price}} * {{ratio}} * {{audioRatio}} * {{audioCompRatio}} = ${{total}} / 1M tokens": "Audio completion: ${{price}} * {{ratio}} * {{audioRatio}} * {{audioCompRatio}} = ${{total}} / 1M tokens",
+  "音频输出：${{price}} * {{ratio}} * {{audioRatio}} * {{audioCompRatio}} = ${{total}} / 1M tokens": "Audio completion: ${{price}} * {{ratio}} * {{audioRatio}} * {{audioCompRatio}} = ${{total}} / 1M tokens",
+  "输入 {{nonImageInput}} tokens + 图片输入 {{imageInput}} tokens * {{imageRatio}} / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}": "Input {{nonImageInput}} tokens + Image input {{imageInput}} tokens * {{imageRatio}} / 1M tokens * ${{price}} + Output {{completion}} tokens / 1M tokens * ${{compPrice}} * Group {{ratio}} = ${{total}}",
   "（文字 + 音频）* 分组倍率 {{ratio}} = ${{total}}": "(Text + Audio) * Group ratio {{ratio}} = ${{total}}",
   "文字提示 {{input}} tokens / 1M tokens * ${{price}} + 文字补全 {{completion}} tokens / 1M tokens * ${{compPrice}} +": "Text prompt {{input}} tokens / 1M tokens * ${{price}} + Text completion {{completion}} tokens / 1M tokens * ${{compPrice}} +",
-  "提示 {{input}} tokens / 1M tokens * ${{price}} + 补全 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}": "Prompt {{input}} tokens / 1M tokens * ${{price}} + Completion {{completion}} tokens / 1M tokens * ${{compPrice}} * Group {{ratio}} = ${{total}}",
+  "输入 {{input}} tokens / 1M tokens * ${{price}} + 输出 {{completion}} tokens / 1M tokens * ${{compPrice}} * 分组 {{ratio}} = ${{total}}": "Prompt {{input}} tokens / 1M tokens * ${{price}} + Completion {{completion}} tokens / 1M tokens * ${{compPrice}} * Group {{ratio}} = ${{total}}",
   "价格：${{price}} * 分组：{{ratio}}": "Price: ${{price}} * Group: {{ratio}}",
   "模型: {{ratio}} * 分组: {{groupRatio}}": "Model: {{ratio}} * Group: {{groupRatio}}",
   "统计额度": "Statistical quota",
@@ -1080,7 +1086,7 @@
   "没有账户？": "No account? ",
   "请输入 AZURE_OPENAI_ENDPOINT，例如：https://docs-test-001.openai.azure.com": "Please enter AZURE_OPENAI_ENDPOINT, e.g.: https://docs-test-001.openai.azure.com",
   "默认 API 版本": "Default API Version",
-  "请输入默认 API 版本，例如：2024-12-01-preview": "Please enter default API version, e.g.: 2024-12-01-preview.",
+  "请输入默认 API 版本，例如：2025-04-01-preview": "Please enter default API version, e.g.: 2025-04-01-preview.",
   "请为渠道命名": "Please name the channel",
   "请选择可以使用该渠道的分组": "Please select groups that can use this channel",
   "请在系统设置页面编辑分组倍率以添加新的分组：": "Please edit Group ratios in system settings to add new groups:",
@@ -1368,4 +1374,4 @@
   "适用于展示系统功能的场景。": "Suitable for scenarios where the system functions are displayed.",
   "可在初始化后修改": "Can be modified after initialization",
   "初始化系统": "Initialize system"
-}
+}

web/src/pages/Channel/EditChannel.js

@@ -25,8 +25,10 @@ import {
   Checkbox,
   Banner,
   Modal,
+  ImagePreview,
 } from '@douyinfe/semi-ui';
 import { getChannelModels, loadChannelModels } from '../../components/utils.js';
+import { IconHelpCircle } from '@douyinfe/semi-icons';
 
 const MODEL_MAPPING_EXAMPLE = {
   'gpt-3.5-turbo': 'gpt-3.5-turbo-0125',
@@ -96,6 +98,8 @@ const EditChannel = (props) => {
   const [basicModels, setBasicModels] = useState([]);
   const [fullModels, setFullModels] = useState([]);
   const [customModel, setCustomModel] = useState('');
+  const [modalImageUrl, setModalImageUrl] = useState('');
+  const [isModalOpenurl, setIsModalOpenurl] = useState(false);
   const handleInputChange = (name, value) => {
     if (name === 'base_url' && value.endsWith('/v1')) {
       Modal.confirm({
@@ -303,7 +307,7 @@ const EditChannel = (props) => {
     fetchModels().then();
     fetchGroups().then();
     if (isEdit) {
-      loadChannel().then(() => { });
+      loadChannel().then(() => {});
     } else {
       setInputs(originInputs);
       let localModels = getChannelModels(inputs.type);
@@ -472,7 +476,30 @@ const EditChannel = (props) => {
               <div style={{ marginTop: 10 }}>
                 <Banner
                   type={'warning'}
-                  description={t('注意，系统请求的时模型名称中的点会被剔除，例如：gpt-4.5-preview会请求为gpt-45-preview，所以部署的模型名称需要去掉点')}
+                  description={
+                    <>
+                      {t(
+                        '2025年5月10日后添加的渠道，不需要再在部署的时候移除模型名称中的"."',
+                      )}
+                      {/*<br />*/}
+                      {/*<Typography.Text*/}
+                      {/*  style={{*/}
+                      {/*    color: 'rgba(var(--semi-blue-5), 1)',*/}
+                      {/*    userSelect: 'none',*/}
+                      {/*    cursor: 'pointer',*/}
+                      {/*  }}*/}
+                      {/*  onClick={() => {*/}
+                      {/*    setModalImageUrl(*/}
+                      {/*      '/azure_model_name.png',*/}
+                      {/*    );*/}
+                      {/*    setIsModalOpenurl(true)*/}
+
+                      {/*  }}*/}
+                      {/*>*/}
+                      {/*  {t('查看示例')}*/}
+                      {/*</Typography.Text>*/}
+                    </>
+                  }
                 ></Banner>
               </div>
               <div style={{ marginTop: 10 }}>
@@ -498,7 +525,7 @@ const EditChannel = (props) => {
               <Input
                 label={t('默认 API 版本')}
                 name='azure_other'
-                placeholder={t('请输入默认 API 版本，例如：2024-12-01-preview')}
+                placeholder={t('请输入默认 API 版本，例如：2025-04-01-preview')}
                 onChange={(value) => {
                   handleInputChange('other', value);
                 }}
@@ -560,25 +587,35 @@ const EditChannel = (props) => {
             value={inputs.name}
             autoComplete='new-password'
           />
-          {inputs.type !== 3 && inputs.type !== 8 && inputs.type !== 22 && inputs.type !== 36 && inputs.type !== 45 && (
-            <>
-              <div style={{ marginTop: 10 }}>
-                <Typography.Text strong>{t('API地址')}：</Typography.Text>
-              </div>
-              <Tooltip content={t('对于官方渠道，new-api已经内置地址，除非是第三方代理站点或者Azure的特殊接入地址，否则不需要填写')}>
-                <Input
-                  label={t('API地址')}
-                  name="base_url"
-                  placeholder={t('此项可选，用于通过自定义API地址来进行 API 调用，末尾不要带/v1和/')}
-                  onChange={(value) => {
-                    handleInputChange('base_url', value);
-                  }}
-                  value={inputs.base_url}
-                  autoComplete="new-password"
-                />
-              </Tooltip>
-            </>
-          )}
+          {inputs.type !== 3 &&
+            inputs.type !== 8 &&
+            inputs.type !== 22 &&
+            inputs.type !== 36 &&
+            inputs.type !== 45 && (
+              <>
+                <div style={{ marginTop: 10 }}>
+                  <Typography.Text strong>{t('API地址')}：</Typography.Text>
+                </div>
+                <Tooltip
+                  content={t(
+                    '对于官方渠道，new-api已经内置地址，除非是第三方代理站点或者Azure的特殊接入地址，否则不需要填写',
+                  )}
+                >
+                  <Input
+                    label={t('API地址')}
+                    name='base_url'
+                    placeholder={t(
+                      '此项可选，用于通过自定义API地址来进行 API 调用，末尾不要带/v1和/',
+                    )}
+                    onChange={(value) => {
+                      handleInputChange('base_url', value);
+                    }}
+                    value={inputs.base_url}
+                    autoComplete='new-password'
+                  />
+                </Tooltip>
+              </>
+            )}
           <div style={{ marginTop: 10 }}>
             <Typography.Text strong>{t('密钥')}：</Typography.Text>
           </div>
@@ -737,10 +774,10 @@ const EditChannel = (props) => {
                 name='other'
                 placeholder={t(
                   '请输入部署地区，例如：us-central1\n支持使用模型映射格式\n' +
-                  '{\n' +
-                  '    "default": "us-central1",\n' +
-                  '    "claude-3-5-sonnet-20240620": "europe-west1"\n' +
-                  '}',
+                    '{\n' +
+                    '    "default": "us-central1",\n' +
+                    '    "claude-3-5-sonnet-20240620": "europe-west1"\n' +
+                    '}',
                 )}
                 autosize={{ minRows: 2 }}
                 onChange={(value) => {
@@ -1109,6 +1146,11 @@ const EditChannel = (props) => {
             {t('填入模板')}
           </Typography.Text>
         </Spin>
+        <ImagePreview
+          src={modalImageUrl}
+          visible={isModalOpenurl}
+          onVisibleChange={(visible) => setIsModalOpenurl(visible)}
+        />
       </SideSheet>
     </>
   );

web/src/pages/Setting/Model/SettingGeminiModel.js

@@ -157,7 +157,6 @@ export default function SettingGeminiModel(props) {
                 <Text>
                   {t(
                     "和Claude不同，默认情况下Gemini的思考模型会自动决定要不要思考，就算不开启适配模型也可以正常使用，" +
-                    "-nothinking后缀（BudgetTokens=0，思考关闭）也会返回少量的思考token，这是gemini的特性，" +
                     "如果您需要计费，推荐设置无后缀模型价格按思考价格设置"
                   )}
                 </Text>

web/src/pages/Setting/RateLimit/SettingsRequestRateLimit.js

@@ -6,6 +6,7 @@ import {
   showError,
   showSuccess,
   showWarning,
+  verifyJSON,
 } from '../../../helpers';
 import { useTranslation } from 'react-i18next';
 
@@ -18,6 +19,7 @@ export default function RequestRateLimit(props) {
     ModelRequestRateLimitCount: -1,
     ModelRequestRateLimitSuccessCount: 1000,
     ModelRequestRateLimitDurationMinutes: 1,
+    ModelRequestRateLimitGroup: '',
   });
   const refForm = useRef();
   const [inputsRow, setInputsRow] = useState(inputs);
@@ -46,6 +48,13 @@ export default function RequestRateLimit(props) {
           if (res.includes(undefined))
             return showError(t('部分保存失败，请重试'));
         }
+
+      for (let i = 0; i < res.length; i++) {
+        if (!res[i].data.success) {
+          return showError(res[i].data.message);
+        }
+      }
+
         showSuccess(t('保存成功'));
         props.refresh();
       })
@@ -147,6 +156,41 @@ export default function RequestRateLimit(props) {
                 />
               </Col>
             </Row>
+            <Row>
+              <Col xs={24} sm={16}>
+                <Form.TextArea
+                  label={t('分组速率限制')}
+                  placeholder={t(
+                    '{\n  "default": [200, 100],\n  "vip": [0, 1000]\n}',
+                  )}
+                  field={'ModelRequestRateLimitGroup'}
+                autosize={{ minRows: 5, maxRows: 15 }}
+                trigger='blur'
+                        stopValidateWithError
+                rules={[
+                  {
+                  validator: (rule, value) => verifyJSON(value),
+                  message: t('不是合法的 JSON 字符串'),
+                  },
+                ]}
+                  extraText={
+                    <div>
+                      <p style={{ marginBottom: -15 }}>{t('说明：')}</p>
+                      <ul>
+                        <li>{t('使用 JSON 对象格式，格式为：{"组名": [最多请求次数, 最多请求完成次数]}')}</li>
+                      <li>{t('示例：{"default": [200, 100], "vip": [0, 1000]}。')}</li>
+                      <li>{t('[最多请求次数]必须大于等于0，[最多请求完成次数]必须大于等于1。')}</li>
+                        <li>{t('分组速率配置优先级高于全局速率限制。')}</li>
+                        <li>{t('限制周期统一使用上方配置的“限制周期”值。')}</li>
+                      </ul>
+                    </div>
+                  }
+                  onChange={(value) => {
+                    setInputs({ ...inputs, ModelRequestRateLimitGroup: value });
+                  }}
+                />
+              </Col>
+            </Row>
             <Row>
               <Button size='default' onClick={onSubmit}>
                 {t('保存模型速率限制')}

web/src/pages/User/EditUser.js

@@ -239,7 +239,7 @@ const EditUser = (props) => {
             readonly
           />
           <div style={{ marginTop: 20 }}>
-            <Typography.Text>{t('`已绑定的 OIDC 账户')}</Typography.Text>
+            <Typography.Text>{t('已绑定的 OIDC 账户')}</Typography.Text>
           </div>
           <Input
             name='oidc_id'