Using Chutes with OpenClaw

Prerequisites

• ✓An existing OpenClaw installation with the `openclaw` CLI available in your shell.
• ✓A Chutes account with access to the model catalog.
• ✓For API-key auth, a Chutes API key from https://chutes.ai/settings/api-keys.
• ✓For OAuth auth, the ability to complete a browser-based login flow from your machine or via a URL + redirect-paste flow on a remote host.

Steps

1. 1

   Understand how the Chutes provider is wired in OpenClaw

   Before you start configuring, you need to know how OpenClaw talks to Chutes. ai/v1`, and it supports both OAuth and API-key auth.

   Both auth methods share the same provider id and register a Chutes catalog plus a default model, so you can swap auth without changing your agent config.

   textCopy

   Provider: `chutes`
   API: OpenAI-compatible
   Base URL: `https://llm.chutes.ai/v1`
   Auth:
       *   OAuth via `openclaw onboard --auth-choice chutes`
       *   API key via `openclaw onboard --auth-choice chutes-api-key`
       *   Runtime env vars: `CHUTES_API_KEY`, `CHUTES_OAUTH_TOKEN`

2. 2

   Onboard Chutes using OAuth

   Use OAuth if you want OpenClaw to manage token refresh for you. The CLI launches a browser flow locally, or prints a URL plus redirect-paste flow on remote or headless servers, and stores the resulting auth profile so tokens auto-refresh.

   This is usually the best option when you do not want to handle raw API keys in your environment.

   bashCopy

   openclaw onboard --auth-choice chutes

3. 3

   Override OAuth client settings when needed

   If you run Chutes in a more controlled environment, you may need to override the default OAuth client configuration. OpenClaw reads a set of Chutes-specific OAuth env vars so you can plug in your own client id, secret, redirect URI, or scopes.

   Set these before running the OAuth onboarding command so the flow uses your custom app.

   textCopy

   CHUTES_CLIENT_ID
   CHUTES_CLIENT_SECRET
   CHUTES_OAUTH_REDIRECT_URI
   CHUTES_OAUTH_SCOPES

4. 4

   Onboard Chutes using an API key instead of OAuth

   If you prefer static credentials or run in an environment where OAuth is awkward, use the API-key onboarding path. 7-TEE`.

   This path is straightforward for CI or non-interactive deployments.

   bashCopy

   openclaw onboard --auth-choice chutes-api-key

5. 5

   Configure your agents to use Chutes models and aliases

   Once auth is in place, point your agents at specific Chutes models or use the default catalog. 7-TEE` as the primary model and assigns human-friendly aliases to a couple of Chutes models.

   You can also rely on the built-in aliases like `chutes-fast`, `chutes-pro`, and `chutes-vision` that OpenClaw registers automatically.

   jsonCopy

   {
     agents: {
       defaults: {
         model: { primary: "chutes/zai-org/GLM-4.7-TEE" },
         models: {
           "chutes/zai-org/GLM-4.7-TEE": { alias: "Chutes GLM 4.7" },
           "chutes/deepseek-ai/DeepSeek-V3.2-TEE": { alias: "Chutes DeepSeek V3.2" },
         },
       },
     },
   }

6. 6

   Rely on discovery and the built-in starter catalog

   With auth configured, OpenClaw queries the Chutes catalog using your credential and registers the discovered models. If that discovery call fails, OpenClaw falls back to a bundled static catalog so onboarding and startup still succeed.

   This fallback includes a set of current Chutes refs you can target directly in your agent config.

   textCopy

   *   `chutes/zai-org/GLM-4.7-TEE`
   *   `chutes/zai-org/GLM-5-TEE`
   *   `chutes/deepseek-ai/DeepSeek-V3.2-TEE`
   *   `chutes/deepseek-ai/DeepSeek-R1-0528-TEE`
   *   `chutes/moonshotai/Kimi-K2.5-TEE`
   *   `chutes/chutesai/Mistral-Small-3.2-24B-Instruct-2506`
   *   `chutes/Qwen/Qwen3-Coder-Next-TEE`
   *   `chutes/openai/gpt-oss-120b-TEE`

Configuration

Troubleshooting

Frequently asked questions