feat: add Anthropic fast mode support

src/agents/fast-mode.ts

@@ -7,6 +7,14 @@ export type FastModeState = {
   source: "session" | "config" | "default";
 };
 
+export function resolveFastModeParam(
+  extraParams: Record<string, unknown> | undefined,
+): boolean | undefined {
+  return normalizeFastMode(
+    (extraParams?.fastMode ?? extraParams?.fast_mode) as string | boolean | null | undefined,
+  );
+}
+
 function resolveConfiguredFastModeRaw(params: {
   cfg: OpenClawConfig | undefined;
   provider: string;

src/agents/pi-embedded-runner-extraparams.live.test.ts

@@ -6,12 +6,16 @@ import { isTruthyEnvValue } from "../infra/env.js";
 import { applyExtraParamsToAgent } from "./pi-embedded-runner.js";
 
 const OPENAI_KEY = process.env.OPENAI_API_KEY ?? "";
+const ANTHROPIC_KEY = process.env.ANTHROPIC_API_KEY ?? "";
 const GEMINI_KEY = process.env.GEMINI_API_KEY ?? "";
 const LIVE = isTruthyEnvValue(process.env.OPENAI_LIVE_TEST) || isTruthyEnvValue(process.env.LIVE);
+const ANTHROPIC_LIVE =
+  isTruthyEnvValue(process.env.ANTHROPIC_LIVE_TEST) || isTruthyEnvValue(process.env.LIVE);
 const GEMINI_LIVE =
   isTruthyEnvValue(process.env.GEMINI_LIVE_TEST) || isTruthyEnvValue(process.env.LIVE);
 
 const describeLive = LIVE && OPENAI_KEY ? describe : describe.skip;
+const describeAnthropicLive = ANTHROPIC_LIVE && ANTHROPIC_KEY ? describe : describe.skip;
 const describeGeminiLive = GEMINI_LIVE && GEMINI_KEY ? describe : describe.skip;
 
 describeLive("pi embedded extra params (live)", () => {
@@ -65,6 +69,79 @@ describeLive("pi embedded extra params (live)", () => {
     // Should respect maxTokens from config (16) — allow a small buffer for provider rounding.
     expect(outputTokens ?? 0).toBeLessThanOrEqual(20);
   }, 30_000);
+
+  it("verifies OpenAI fast-mode service_tier semantics against the live API", async () => {
+    const headers = {
+      "content-type": "application/json",
+      authorization: `Bearer ${OPENAI_KEY}`,
+    };
+
+    const runProbe = async (serviceTier: "default" | "priority") => {
+      const res = await fetch("https://api.openai.com/v1/responses", {
+        method: "POST",
+        headers,
+        body: JSON.stringify({
+          model: "gpt-5.4",
+          input: "Reply with OK.",
+          max_output_tokens: 32,
+          service_tier: serviceTier,
+        }),
+      });
+      const json = (await res.json()) as {
+        error?: { message?: string };
+        service_tier?: string;
+        status?: string;
+      };
+      expect(res.ok, json.error?.message ?? `HTTP ${res.status}`).toBe(true);
+      return json;
+    };
+
+    const standard = await runProbe("default");
+    expect(standard.service_tier).toBe("default");
+    expect(standard.status).toBe("completed");
+
+    const fast = await runProbe("priority");
+    expect(fast.service_tier).toBe("priority");
+    expect(fast.status).toBe("completed");
+  }, 45_000);
+});
+
+describeAnthropicLive("pi embedded extra params (anthropic live)", () => {
+  it("verifies Anthropic fast-mode service_tier semantics against the live API", async () => {
+    const headers = {
+      "content-type": "application/json",
+      "x-api-key": ANTHROPIC_KEY,
+      "anthropic-version": "2023-06-01",
+    };
+
+    const runProbe = async (serviceTier: "auto" | "standard_only") => {
+      const res = await fetch("https://api.anthropic.com/v1/messages", {
+        method: "POST",
+        headers,
+        body: JSON.stringify({
+          model: "claude-sonnet-4-5",
+          max_tokens: 32,
+          service_tier: serviceTier,
+          messages: [{ role: "user", content: "Reply with OK." }],
+        }),
+      });
+      const json = (await res.json()) as {
+        error?: { message?: string };
+        stop_reason?: string;
+        usage?: { service_tier?: string };
+      };
+      expect(res.ok, json.error?.message ?? `HTTP ${res.status}`).toBe(true);
+      return json;
+    };
+
+    const standard = await runProbe("standard_only");
+    expect(standard.usage?.service_tier).toBe("standard");
+    expect(standard.stop_reason).toBe("end_turn");
+
+    const fast = await runProbe("auto");
+    expect(["standard", "priority"]).toContain(fast.usage?.service_tier);
+    expect(fast.stop_reason).toBe("end_turn");
+  }, 45_000);
 });
 
 describeGeminiLive("pi embedded extra params (gemini live)", () => {

src/agents/pi-embedded-runner-extraparams.test.ts

@@ -201,7 +201,8 @@ describe("applyExtraParamsToAgent", () => {
     model:
       | Model<"openai-responses">
       | Model<"openai-codex-responses">
-      | Model<"openai-completions">;
+      | Model<"openai-completions">
+      | Model<"anthropic-messages">;
     options?: SimpleStreamOptions;
     cfg?: Record<string, unknown>;
     extraParamsOverride?: Record<string, unknown>;
@@ -1683,6 +1684,91 @@ describe("applyExtraParamsToAgent", () => {
     expect(payload.service_tier).toBe("default");
   });
 
+  it("injects service_tier=auto for Anthropic fast mode on direct API-key models", () => {
+    const payload = runResponsesPayloadMutationCase({
+      applyProvider: "anthropic",
+      applyModelId: "claude-sonnet-4-5",
+      extraParamsOverride: { fastMode: true },
+      model: {
+        api: "anthropic-messages",
+        provider: "anthropic",
+        id: "claude-sonnet-4-5",
+        baseUrl: "https://api.anthropic.com",
+      } as unknown as Model<"anthropic-messages">,
+      payload: {},
+    });
+    expect(payload.service_tier).toBe("auto");
+  });
+
+  it("injects service_tier=standard_only for Anthropic fast mode off", () => {
+    const payload = runResponsesPayloadMutationCase({
+      applyProvider: "anthropic",
+      applyModelId: "claude-sonnet-4-5",
+      extraParamsOverride: { fastMode: false },
+      model: {
+        api: "anthropic-messages",
+        provider: "anthropic",
+        id: "claude-sonnet-4-5",
+        baseUrl: "https://api.anthropic.com",
+      } as unknown as Model<"anthropic-messages">,
+      payload: {},
+    });
+    expect(payload.service_tier).toBe("standard_only");
+  });
+
+  it("preserves caller-provided Anthropic service_tier values", () => {
+    const payload = runResponsesPayloadMutationCase({
+      applyProvider: "anthropic",
+      applyModelId: "claude-sonnet-4-5",
+      extraParamsOverride: { fastMode: true },
+      model: {
+        api: "anthropic-messages",
+        provider: "anthropic",
+        id: "claude-sonnet-4-5",
+        baseUrl: "https://api.anthropic.com",
+      } as unknown as Model<"anthropic-messages">,
+      payload: {
+        service_tier: "standard_only",
+      },
+    });
+    expect(payload.service_tier).toBe("standard_only");
+  });
+
+  it("does not inject Anthropic fast mode service_tier for OAuth auth", () => {
+    const payload = runResponsesPayloadMutationCase({
+      applyProvider: "anthropic",
+      applyModelId: "claude-sonnet-4-5",
+      extraParamsOverride: { fastMode: true },
+      model: {
+        api: "anthropic-messages",
+        provider: "anthropic",
+        id: "claude-sonnet-4-5",
+        baseUrl: "https://api.anthropic.com",
+      } as unknown as Model<"anthropic-messages">,
+      options: {
+        apiKey: "sk-ant-oat-test-token",
+      },
+      payload: {},
+    });
+    expect(payload).not.toHaveProperty("service_tier");
+  });
+
+  it("does not inject Anthropic fast mode service_tier for proxied base URLs", () => {
+    const payload = runResponsesPayloadMutationCase({
+      applyProvider: "anthropic",
+      applyModelId: "claude-sonnet-4-5",
+      extraParamsOverride: { fastMode: true },
+      model: {
+        api: "anthropic-messages",
+        provider: "anthropic",
+        id: "claude-sonnet-4-5",
+        baseUrl: "https://proxy.example.com/anthropic",
+      } as unknown as Model<"anthropic-messages">,
+      payload: {},
+    });
+    expect(payload).not.toHaveProperty("service_tier");
+  });
+
   it("applies fast-mode defaults for openai-codex responses without service_tier", () => {
     const payload = runResponsesPayloadMutationCase({
       applyProvider: "openai-codex",

src/agents/pi-embedded-runner/anthropic-stream-wrappers.ts

@@ -1,5 +1,6 @@
 import type { StreamFn } from "@mariozechner/pi-agent-core";
 import { streamSimple } from "@mariozechner/pi-ai";
+import { resolveFastModeParam } from "../fast-mode.js";
 import {
   requiresOpenAiCompatibleAnthropicToolPayload,
   usesOpenAiFunctionAnthropicToolSchema,
@@ -18,6 +19,7 @@ const PI_AI_OAUTH_ANTHROPIC_BETAS = [
   "oauth-2025-04-20",
   ...PI_AI_DEFAULT_ANTHROPIC_BETAS,
 ] as const;
+type AnthropicServiceTier = "auto" | "standard_only";
 
 type CacheRetention = "none" | "short" | "long";
 
@@ -53,6 +55,25 @@ function isAnthropicOAuthApiKey(apiKey: unknown): boolean {
   return typeof apiKey === "string" && apiKey.includes("sk-ant-oat");
 }
 
+function isAnthropicPublicApiBaseUrl(baseUrl: unknown): boolean {
+  if (baseUrl == null) {
+    return true;
+  }
+  if (typeof baseUrl !== "string" || !baseUrl.trim()) {
+    return true;
+  }
+
+  try {
+    return new URL(baseUrl).hostname.toLowerCase() === "api.anthropic.com";
+  } catch {
+    return baseUrl.toLowerCase().includes("api.anthropic.com");
+  }
+}
+
+function resolveAnthropicFastServiceTier(enabled: boolean): AnthropicServiceTier {
+  return enabled ? "auto" : "standard_only";
+}
+
 function requiresAnthropicToolPayloadCompatibilityForModel(model: {
   api?: unknown;
   provider?: unknown;
@@ -304,6 +325,44 @@ export function createAnthropicToolPayloadCompatibilityWrapper(
   };
 }
 
+export function createAnthropicFastModeWrapper(
+  baseStreamFn: StreamFn | undefined,
+  enabled: boolean,
+): StreamFn {
+  const underlying = baseStreamFn ?? streamSimple;
+  const serviceTier = resolveAnthropicFastServiceTier(enabled);
+  return (model, context, options) => {
+    if (
+      model.api !== "anthropic-messages" ||
+      model.provider !== "anthropic" ||
+      !isAnthropicPublicApiBaseUrl(model.baseUrl) ||
+      isAnthropicOAuthApiKey(options?.apiKey)
+    ) {
+      return underlying(model, context, options);
+    }
+
+    const originalOnPayload = options?.onPayload;
+    return underlying(model, context, {
+      ...options,
+      onPayload: (payload) => {
+        if (payload && typeof payload === "object") {
+          const payloadObj = payload as Record<string, unknown>;
+          if (payloadObj.service_tier === undefined) {
+            payloadObj.service_tier = serviceTier;
+          }
+        }
+        return originalOnPayload?.(payload, model);
+      },
+    });
+  };
+}
+
+export function resolveAnthropicFastMode(
+  extraParams: Record<string, unknown> | undefined,
+): boolean | undefined {
+  return resolveFastModeParam(extraParams);
+}
+
 export function createBedrockNoCacheWrapper(baseStreamFn: StreamFn | undefined): StreamFn {
   const underlying = baseStreamFn ?? streamSimple;
   return (model, context, options) =>

src/agents/pi-embedded-runner/extra-params.ts

@@ -5,9 +5,11 @@ import type { ThinkLevel } from "../../auto-reply/thinking.js";
 import type { OpenClawConfig } from "../../config/config.js";
 import {
   createAnthropicBetaHeadersWrapper,
+  createAnthropicFastModeWrapper,
   createAnthropicToolPayloadCompatibilityWrapper,
   createBedrockNoCacheWrapper,
   isAnthropicBedrockModel,
+  resolveAnthropicFastMode,
   resolveAnthropicBetas,
   resolveCacheRetention,
 } from "./anthropic-stream-wrappers.js";
@@ -439,6 +441,12 @@ export function applyExtraParamsToAgent(
   // upstream model-ID heuristics for Gemini 3.1 variants.
   agent.streamFn = createGoogleThinkingPayloadWrapper(agent.streamFn, thinkingLevel);
 
+  const anthropicFastMode = resolveAnthropicFastMode(merged);
+  if (anthropicFastMode !== undefined) {
+    log.debug(`applying Anthropic fast mode=${anthropicFastMode} for ${provider}/${modelId}`);
+    agent.streamFn = createAnthropicFastModeWrapper(agent.streamFn, anthropicFastMode);
+  }
+
   const openAIFastMode = resolveOpenAIFastMode(merged);
   if (openAIFastMode) {
     log.debug(`applying OpenAI fast mode for ${provider}/${modelId}`);