Skip to content

OpenClaw [BETA]

Put Chutes behind OpenClaw channel agents.

OpenClaw is an open-source, self-hosted AI assistant whose gateway connects agents to chat channels such as Discord, Slack, Telegram, and WhatsApp. It can point at Chutes through an OpenAI-compatible provider block. This page keeps the toolkit's [BETA] boundary visible: provider fields are doc-derived until an end-to-end OpenClaw install is exercised.
One-command installerOpenClaw READMEBasic config sourceRouting config source
Status
[BETA] doc-derived
Provider base URL
https://llm.chutes.ai/v1
Live model example
Qwen/Qwen3.8-27B-TEE
Catalog checked
Nov 8, 2026

Full setup walkthrough

  1. 01

    Export the API key

    Set CHUTES_API_KEY=cpk_... in the OpenClaw runtime environment. It is sent as Authorization: Bearer.
    bash
    export CHUTES_API_KEY=cpk_...
  2. 02

    Add Chutes as a provider

    Add an OpenAI-compatible provider block with a live model id pulled from the catalog:
    models.providers.chutes
    json5
    // JSON5
    {
      agents: {
        defaults: {
          model: { primary: "chutes/Qwen/Qwen3.8-27B-TEE" },
        },
      },
      models: {
        providers: {
          chutes: {
            baseUrl: "https://llm.chutes.ai/v1",
            apiKey: "${CHUTES_API_KEY}",
            api: "openai-completions",
            timeoutSeconds: 300,
            models: [
              {
                id: "Qwen/Qwen3.8-27B-TEE",
                name: "Live Chutes model",
                input: ["text"],
                contextWindow: 262144,
              },
            ],
          },
        },
      },
    }
    This integration is [BETA] doc-derived: the Chutes side is verified, but the OpenClaw field names come from docs and have not been exercised end to end. Verify against your OpenClaw build before production.
  3. 03

    Select a model or routing pool

    Use a concrete id for zero setup, or an inline pool like id1,id2:latency. Saved default:* aliases need a dashboard Model Routing pool first.

Verification boundary

Known
Chutes exposes the OpenAI-compatible base URL and Bearer auth OpenClaw provider blocks need.
Source
Toolkit OpenClaw README plus config examples.
Not claimed
A full channel-agent deployment was live-verified from this page.
Required label
[BETA] doc-derived

Common questions

Is there a one-command OpenClaw installer for Chutes?

Yes. On macOS or Linux run curl -fsSL https://chutes.ai/openclaw/init | sh; on Windows PowerShell run irm https://chutes.ai/openclaw/init.ps1 | iex. The installer sets up OpenClaw, adds Chutes as its model provider, and starts the OpenClaw gateway on localhost port 18789. See chutes.ai/openclaw.

Is the OpenClaw integration fully live-verified?

No. The Chutes provider shape is source-backed, but the page preserves the [BETA] doc-derived boundary until an end-to-end OpenClaw install is exercised.

Troubleshooting

Symptom
The provider loads but model calls fail
Likely cause
A stale model ID or non-Bearer key reached inference.
Fix
Refresh /v1/models and export CHUTES_API_KEY=cpk_... before starting OpenClaw.
Symptom
A config includes static pricing
Likely cause
The toolkit example marks pricing as a snapshot.
Fix
Render current prices from /v1/models instead of copying them into page copy.
Symptom
The page sounds fully verified
Likely cause
The OpenClaw field mapping has not been exercised end to end here.
Fix
Keep the [BETA] doc-derived label visible.