github/openclaw
1
0
Fork 0
mirror of https://github.com/openclaw/openclaw.git synced 2026-05-11 00:14:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(ollama): unify context window handling across discovery, merge, and OpenAI-compat transport (#29205)

Browse Source 
* fix(ollama): inject num_ctx for OpenAI-compatible transport

* fix(ollama): discover per-model context and preserve higher limits

* fix(agents): prefer matching provider model for fallback limits

* fix(types): require numeric token limits in provider model merge

* fix(types): accept unknown payload in ollama num_ctx wrapper

* fix(types): simplify ollama settled-result extraction

* config(models): add provider flag for Ollama OpenAI num_ctx injection

* config(schema): allow provider num_ctx injection flag

* config(labels): label provider num_ctx injection flag

* config(help): document provider num_ctx injection flag

* agents(ollama): gate OpenAI num_ctx injection with provider config

* tests(ollama): cover provider num_ctx injection flag behavior

* docs(config): list provider num_ctx injection option

* docs(ollama): document OpenAI num_ctx injection toggle

* docs(config): clarify merge token-limit precedence

* config(help): note merge uses higher model token limits

* fix(ollama): cap /api/show discovery concurrency

* fix(ollama): restrict num_ctx injection to OpenAI compat

* tests(ollama): cover ipv6 and compat num_ctx gating

* fix(ollama): detect remote compat endpoints for ollama-labeled providers

* fix(ollama): cap per-model /api/show lookups to bound discovery load
This commit is contained in:
Vincent Koc
2026-02-27 17:20:47 -08:00
committed by GitHub
parent 70a4f25ab1
commit f16ecd1dac
14 changed files with 582 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

77
src/agents/models-config.providers.ts
View File

@@ -144,6 +144,8 @@ const QWEN_PORTAL_DEFAULT_COST = {
const OLLAMA_BASE_URL = OLLAMA_NATIVE_BASE_URL;
const OLLAMA_API_BASE_URL = OLLAMA_BASE_URL;
const OLLAMA_SHOW_CONCURRENCY = 8;
const OLLAMA_SHOW_MAX_MODELS = 200;
const OLLAMA_DEFAULT_CONTEXT_WINDOW = 128000;
const OLLAMA_DEFAULT_MAX_TOKENS = 8192;
const OLLAMA_DEFAULT_COST = {
@@ -236,6 +238,38 @@ export function resolveOllamaApiBase(configuredBaseUrl?: string): string {
return trimmed.replace(/\/v1$/i, "");
}
async function queryOllamaContextWindow(
apiBase: string,
modelName: string,
): Promise<number | undefined> {
try {
const response = await fetch(`${apiBase}/api/show`, {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({ name: modelName }),
signal: AbortSignal.timeout(3000),
});
if (!response.ok) {
return undefined;
}
const data = (await response.json()) as { model_info?: Record<string, unknown> };
if (!data.model_info) {
return undefined;
}
for (const [key, value] of Object.entries(data.model_info)) {
if (key.endsWith(".context_length") && typeof value === "number" && Number.isFinite(value)) {
const contextWindow = Math.floor(value);
if (contextWindow > 0) {
return contextWindow;
}
}
}
return undefined;
} catch {
return undefined;
}
}
async function discoverOllamaModels(
baseUrl?: string,
opts?: { quiet?: boolean },
@@ -260,20 +294,35 @@ async function discoverOllamaModels(
log.debug("No Ollama models found on local instance");
return [];
}
return data.models.map((model) => {
const modelId = model.name;
const isReasoning =
modelId.toLowerCase().includes("r1") || modelId.toLowerCase().includes("reasoning");
return {
id: modelId,
name: modelId,
reasoning: isReasoning,
input: ["text"],
cost: OLLAMA_DEFAULT_COST,
contextWindow: OLLAMA_DEFAULT_CONTEXT_WINDOW,
maxTokens: OLLAMA_DEFAULT_MAX_TOKENS,
};
});
const modelsToInspect = data.models.slice(0, OLLAMA_SHOW_MAX_MODELS);
if (modelsToInspect.length < data.models.length && !opts?.quiet) {
log.warn(
`Capping Ollama /api/show inspection to ${OLLAMA_SHOW_MAX_MODELS} models (received ${data.models.length})`,
);
}
const discovered: ModelDefinitionConfig[] = [];
for (let index = 0; index < modelsToInspect.length; index += OLLAMA_SHOW_CONCURRENCY) {
const batch = modelsToInspect.slice(index, index + OLLAMA_SHOW_CONCURRENCY);
const batchDiscovered = await Promise.all(
batch.map(async (model) => {
const modelId = model.name;
const contextWindow = await queryOllamaContextWindow(apiBase, modelId);
const isReasoning =
modelId.toLowerCase().includes("r1") || modelId.toLowerCase().includes("reasoning");
return {
id: modelId,
name: modelId,
reasoning: isReasoning,
input: ["text"],
cost: OLLAMA_DEFAULT_COST,
contextWindow: contextWindow ?? OLLAMA_DEFAULT_CONTEXT_WINDOW,
maxTokens: OLLAMA_DEFAULT_MAX_TOKENS,
} satisfies ModelDefinitionConfig;
}),
);
discovered.push(...batchDiscovered);
}
return discovered;
} catch (error) {
if (!opts?.quiet) {
log.warn(`Failed to discover Ollama models: ${String(error)}`);