clawdbot/docs/gateway/local-models.md

summary, read_when

summary	read_when
Run Clawdbot on local LLMs (LM Studio, vLLM, LiteLLM, custom OpenAI endpoints)	You want to serve models from your own GPU box You are wiring LM Studio or an OpenAI-compatible proxy You need the safest local model guidance

Local models

Local is doable, but Clawdbot expects large context + strong defenses against prompt injection. Small cards truncate context and leak safety. Aim high: ≥2 maxed-out Mac Studios or equivalent GPU rig (~$30k+). A single 24 GB GPU works only for lighter prompts with higher latency. Use the largest / full-size model variant you can run; aggressively quantized or “small” checkpoints raise prompt-injection risk (see Security).

Recommended: LM Studio + MiniMax M2.1 (Responses API, full-size)

Best current local stack. Load MiniMax M2.1 in LM Studio, enable the local server (default http://127.0.0.1:1234), and use Responses API to keep reasoning separate from final text.

{
  agents: {
    defaults: {
      model: { primary: "lmstudio/minimax-m2.1-gs32" },
      models: {
        "anthropic/claude-opus-4-5": { alias: "Opus" },
        "lmstudio/minimax-m2.1-gs32": { alias: "Minimax" }
      }
    }
  },
  models: {
    mode: "merge",
    providers: {
      lmstudio: {
        baseUrl: "http://127.0.0.1:1234/v1",
        apiKey: "lmstudio",
        api: "openai-responses",
        models: [
          {
            id: "minimax-m2.1-gs32",
            name: "MiniMax M2.1 GS32",
            reasoning: false,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 196608,
            maxTokens: 8192
          }
        ]
      }
    }
  }
}

Setup checklist

• Install LM Studio: https://lmstudio.ai
• In LM Studio, download the largest MiniMax M2.1 build available (avoid “small”/heavily quantized variants), start the server, confirm http://127.0.0.1:1234/v1/models lists it.
• Keep the model loaded; cold-load adds startup latency.
• Adjust contextWindow/maxTokens if your LM Studio build differs.
• For WhatsApp, stick to Responses API so only final text is sent.

Model catalog (remote + local)

Tier	Model ID	Context	Who downloads	Notes
Hosted	anthropic/claude-opus-4-5	200k	Provider	Latest Claude; keep as high-quality fallback.
Hosted	anthropic/claude-sonnet-4-5	200k	Provider	Cheaper Claude; good default.
Hosted	openai/gpt-4.1	128k	Provider	Latest GPT-4; strong tools + reasoning.
Hosted	openai/gpt-4.1-mini	128k	Provider	Fast/cheap GPT-4 family; good fallback.
Local	lmstudio/minimax-m2.1-gs32	~196k (build-dependent)	You (LM Studio UI)	Recommended local heavy model; keep loaded.
Local	Custom vllm / litellm model	server-defined	You (server)	Any OpenAI-compatible endpoint; align context settings.

Keep hosted models configured even when running local; use models.mode: "merge" so fallbacks stay available.

Hybrid config: hosted primary, local fallback

{
  agents: {
    defaults: {
      model: {
        primary: "anthropic/claude-sonnet-4-5",
        fallbacks: ["lmstudio/minimax-m2.1-gs32", "openai/gpt-4.1-mini"]
      },
      models: {
        "anthropic/claude-sonnet-4-5": { alias: "Sonnet" },
        "lmstudio/minimax-m2.1-gs32": { alias: "MiniMax Local" },
        "openai/gpt-4.1-mini": { alias: "GPT-4.1 mini" }
      }
    }
  },
  models: {
    mode: "merge",
    providers: {
      lmstudio: {
        baseUrl: "http://127.0.0.1:1234/v1",
        apiKey: "lmstudio",
        api: "openai-responses",
        models: [
          {
            id: "minimax-m2.1-gs32",
            name: "MiniMax M2.1 GS32",
            reasoning: false,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 196608,
            maxTokens: 8192
          }
        ]
      }
    }
  }
}

Local-first with hosted safety net

Swap the primary and fallback order; keep the same providers block and models.mode: "merge" so you can fall back to Sonnet or GPT-4.1 when the local box is down.

Regional hosting / data routing

• Hosted MiniMax/Kimi/GLM variants also exist on OpenRouter with region-pinned endpoints (e.g., US-hosted). Pick the regional variant there to keep traffic in your chosen jurisdiction while still using models.mode: "merge" for Anthropic/OpenAI fallbacks.
• Local-only remains the strongest privacy path; hosted regional routing is the middle ground when you need provider features but want control over data flow.

Other OpenAI-compatible local proxies

vLLM, LiteLLM, OAI-proxy, or custom gateways work if they expose an OpenAI-style /v1 endpoint. Replace the provider block above with your endpoint and model ID:

{
  models: {
    mode: "merge",
    providers: {
      local: {
        baseUrl: "http://127.0.0.1:8000/v1",
        apiKey: "sk-local",
        api: "openai-responses",
        models: [
          {
            id: "my-local-model",
            name: "Local Model",
            reasoning: false,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 120000,
            maxTokens: 8192
          }
        ]
      }
    }
  }
}

Keep models.mode: "merge" so hosted models stay available as fallbacks.

Troubleshooting

• Gateway can reach the proxy? curl http://127.0.0.1:1234/v1/models.
• LM Studio model unloaded? Reload; cold start is a common “hanging” cause.
• Context errors? Lower contextWindow or raise your server limit.
• Safety: local models skip provider-side filters; keep agents narrow and compaction on to limit prompt injection blast radius.