/**
* Ollama Backend — talks to a locally running Ollama instance.
*
* Ollama runs models like Llama 3, Mistral, CodeGemma, Phi, etc.
* on your own hardware. No subscriptions, no API keys, no rate limits.
* Just you and your GPU.
*
* Default endpoint: http://localhost:11434
* Ollama exposes an OpenAI-compatible API at /v1/chat/completions
* so the implementation looks almost identical to CLIProxyAPI.
*
* Why this matters for the project:
* - HydraMCP isn't just a CLIProxyAPI wrapper anymore
* - Users can mix cloud models (via CLIProxyAPI) with local models
* - "Compare GPT-4o vs my local Llama 3" becomes possible
* - Zero cost for the local side
*/
import { Provider, ModelInfo, QueryOptions, QueryResponse } from "./provider.js";
export class OllamaProvider implements Provider {
name = "Ollama";
private baseUrl: string;
constructor(baseUrl?: string) {
this.baseUrl = baseUrl ?? process.env.OLLAMA_URL ?? "http://localhost:11434";
}
async healthCheck(): Promise<boolean> {
try {
const res = await fetch(`${this.baseUrl}/api/tags`);
return res.ok;
} catch {
return false;
}
}
async listModels(): Promise<ModelInfo[]> {
const res = await fetch(`${this.baseUrl}/api/tags`);
if (!res.ok) {
throw new Error(`Failed to list Ollama models: ${res.status}`);
}
const data = (await res.json()) as {
models?: Array<{ name: string; details?: { family?: string } }>;
};
return (data.models ?? []).map((m) => ({
id: m.name,
name: m.name,
provider: "ollama",
}));
}
async query(
model: string,
prompt: string,
options?: QueryOptions
): Promise<QueryResponse> {
const startTime = Date.now();
// Ollama supports OpenAI-compatible endpoint
const body: Record<string, unknown> = {
model,
messages: [
...(options?.system_prompt
? [{ role: "system", content: options.system_prompt }]
: []),
{ role: "user", content: prompt },
],
stream: false,
};
if (options?.temperature !== undefined) {
body.options = { temperature: options.temperature };
}
// Use Ollama's native chat endpoint (more reliable than /v1 compat)
const res = await fetch(`${this.baseUrl}/api/chat`, {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify(body),
});
if (!res.ok) {
const errorText = await res.text();
throw new Error(`Ollama query failed (${res.status}): ${errorText}`);
}
const data = (await res.json()) as {
message?: { content?: string };
done_reason?: string;
prompt_eval_count?: number;
eval_count?: number;
};
const latency_ms = Date.now() - startTime;
const prompt_tokens = data.prompt_eval_count ?? 0;
const completion_tokens = data.eval_count ?? 0;
return {
model,
content: data.message?.content ?? "",
usage: {
prompt_tokens,
completion_tokens,
total_tokens: prompt_tokens + completion_tokens,
},
latency_ms,
finish_reason: data.done_reason ?? "stop",
};
}
}
