> ## Documentation Index > Fetch the complete documentation index at: https://docs.anyapi.ai/llms.txt > Use this file to discover all available pages before exploring further. # OpenClaw > Connect AnyAPI as a custom provider in OpenClaw for multi-model AI agent workflows # OpenClaw [OpenClaw](https://github.com/openclaw/openclaw) is an AI coding agent with support for custom OpenAI-compatible providers, fallback chains, and messaging gateway mode (Discord, Slack, Telegram, etc.). AnyAPI is not a bundled provider in OpenClaw. You wire it in as a **custom OpenAI-compatible provider** under `models.providers`. Once registered, OpenClaw addresses models as `anyapi//` (your provider ID prefix + the upstream model ID). OpenClaw uses **JSON5** at `~/.openclaw/openclaw.json` — unquoted keys, trailing commas, and `//` comments are all valid. ## Quick Start The fastest path is the non-interactive onboarding wizard: ```bash theme={"system"} ANYAPI_API_KEY='your-key-here' openclaw onboard --non-interactive \ --auth-choice custom-api-key \ --custom-base-url "https://api.anyapi.ai/v1" \ --custom-model-id "openai/gpt-4o" \ --secret-input-mode ref \ --custom-compatibility openai \ --accept-risk ``` This registers a custom provider with `baseUrl: https://api.anyapi.ai/v1`, stores the API key as an env reference, and sets `openai/gpt-4o` as the default model. Then start OpenClaw: ```bash theme={"system"} openclaw ``` Prefer interactive setup? Run `openclaw onboard` without flags to walk through provider, URL, key, and model selection step by step. ## Manual Configuration Edit `~/.openclaw/openclaw.json` directly for full control: ```json5 theme={"system"} { env: { ANYAPI_API_KEY: "your-key-here", }, models: { providers: { anyapi: { baseUrl: "https://api.anyapi.ai/v1", apiKey: { source: "env", provider: "default", id: "ANYAPI_API_KEY", }, api: "openai-completions", models: [ { id: "openai/gpt-4o" }, { id: "openai/gpt-4o-mini" }, { id: "anthropic/claude-sonnet-4.6" }, { id: "google/gemini-2.5-pro" }, ], }, }, }, agents: { defaults: { model: { primary: "anyapi/openai/gpt-4o", fallbacks: [ "anyapi/anthropic/claude-sonnet-4.6", "anyapi/openai/gpt-4o-mini", ], }, }, }, } ``` Restart OpenClaw for changes to take effect: ```bash theme={"system"} openclaw gateway restart # if running as gateway daemon ``` ## Model Reference Format OpenClaw composes model refs as `/`. With provider ID `anyapi`: * `anyapi/openai/gpt-4o` — correct * `anyapi/anthropic/claude-sonnet-4.6` — correct * `openai/gpt-4o` — won't route through the `anyapi` provider * `gpt-4o` — missing vendor prefix, AnyAPI returns HTTP 400 The `id` field inside `models.providers.anyapi.models[]` is the **upstream** ID sent to AnyAPI (e.g. `openai/gpt-4o`). The `anyapi/` prefix is added by OpenClaw based on your provider ID. Browse all available models at [anyapi.ai/ai-models](https://anyapi.ai/ai-models) or query the API: ```bash theme={"system"} curl -sS -H "Authorization: Bearer $ANYAPI_API_KEY" \ https://api.anyapi.ai/v1/models | jq -r '.data[].id' ``` ## Configuration Examples ```json5 theme={"system"} { models: { providers: { anyapi: { baseUrl: "https://api.anyapi.ai/v1", apiKey: { source: "env", provider: "default", id: "ANYAPI_API_KEY" }, models: [{ id: "openai/gpt-4o" }], }, }, }, agents: { defaults: { model: { primary: "anyapi/openai/gpt-4o" }, }, }, } ``` ```json5 theme={"system"} { models: { providers: { anyapi: { baseUrl: "https://api.anyapi.ai/v1", apiKey: { source: "env", provider: "default", id: "ANYAPI_API_KEY" }, models: [{ id: "anthropic/claude-sonnet-4.6" }], }, }, }, agents: { defaults: { model: { primary: "anyapi/anthropic/claude-sonnet-4.6" }, }, }, } ``` ```json5 theme={"system"} { models: { providers: { anyapi: { baseUrl: "https://api.anyapi.ai/v1", apiKey: { source: "env", provider: "default", id: "ANYAPI_API_KEY" }, models: [ { id: "openai/gpt-4o" }, { id: "anthropic/claude-sonnet-4.6" }, { id: "openai/gpt-4o-mini" }, ], }, }, }, agents: { defaults: { model: { primary: "anyapi/openai/gpt-4o", fallbacks: [ "anyapi/anthropic/claude-sonnet-4.6", "anyapi/openai/gpt-4o-mini", ], timeoutMs: 120000, }, }, }, } ``` ## Advanced Configuration Set per-model defaults via `agents.defaults.models`: ```json5 theme={"system"} { agents: { defaults: { models: { "anyapi/openai/gpt-4o": { params: { temperature: 0.7, max_tokens: 4096, }, }, "anyapi/openai/o3": { params: { max_completion_tokens: 8192, // o-series uses max_completion_tokens }, }, }, }, }, } ``` Enrich model entries to enable image inputs and set context windows: ```json5 theme={"system"} models: [ { id: "openai/gpt-4o", name: "GPT-4o (via AnyAPI)", input: ["text", "image"], contextWindow: 128000, maxTokens: 16384, }, { id: "google/gemini-2.5-pro", input: ["text", "image"], contextWindow: 1000000, reasoning: true, }, ] ``` OpenClaw won't infer image capability for custom providers automatically — declare it explicitly. Add request headers for tracking: ```json5 theme={"system"} anyapi: { baseUrl: "https://api.anyapi.ai/v1", apiKey: { source: "env", provider: "default", id: "ANYAPI_API_KEY" }, headers: { "X-My-App": "my-app-v1", }, models: [{ id: "openai/gpt-4o" }], } ``` For multi-account setups, OpenClaw supports auth profiles. Store a key in OS keychain: ```bash theme={"system"} openclaw models auth add --provider anyapi ``` ## CLI Reference ```bash theme={"system"} # Onboarding openclaw onboard # interactive setup openclaw onboard --non-interactive ... # scripted setup # Model management openclaw models list # list configured models openclaw models set anyapi/openai/gpt-4o # set default model # Config openclaw config validate # validate config openclaw config get models.providers.anyapi.baseUrl openclaw config set agents.defaults.model.primary "anyapi/openai/gpt-4o" # Gateway openclaw gateway status openclaw gateway restart openclaw gateway logs ``` ## Troubleshooting OpenClaw rejects unknown keys. Use the canonical field names: * `baseUrl` (camelCase, not `baseURL`) * `apiKey` (object or `${VAR}` template, not `apiKeyEnvVar`) Validate with `openclaw config validate`. The `apiKey` reference doesn't resolve. 1. Confirm the env var is set: `echo $ANYAPI_API_KEY` 2. Confirm the `apiKey` block points to it: ```json5 theme={"system"} apiKey: { source: "env", provider: "default", id: "ANYAPI_API_KEY" } ``` 3. Restart OpenClaw — env changes don't apply to a running daemon. Your key is invalid, expired, or revoked. 1. Validate the key at [dash.anyapi.ai](https://dash.anyapi.ai) 2. Test directly: ```bash theme={"system"} curl https://api.anyapi.ai/v1/chat/completions \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $ANYAPI_API_KEY" \ -d '{"model":"openai/gpt-4o-mini","messages":[{"role":"user","content":"test"}],"max_tokens":10}' ``` The model ID is unknown or missing the vendor prefix. 1. List the catalog: `curl -H "Authorization: Bearer $ANYAPI_API_KEY" https://api.anyapi.ai/v1/models` 2. Use `vendor/model` form (e.g. `openai/gpt-4o`, not bare `gpt-4o`) 3. Confirm the model is listed under `models.providers.anyapi.models[]` in your config Set `baseUrl` to exactly `https://api.anyapi.ai/v1` — no trailing slash, no `/chat/completions` suffix (OpenClaw appends that). Check usage at [dash.anyapi.ai](https://dash.anyapi.ai), upgrade your plan, or add fallbacks to spread load. ## Best Practices 1. **Store keys as env refs, not plaintext** — prefer `secret-input-mode ref` during onboarding 2. **Configure fallbacks** across different vendors for cross-provider resilience 3. **Set `timeoutMs` for slow models** — reasoning models like `openai/o3` can take minutes 4. **Pin model capability metadata** — declare `input: ["text", "image"]` explicitly for vision models 5. **Always use `vendor/model` format** — bare names return HTTP 400 6. **Monitor usage** at [dash.anyapi.ai](https://dash.anyapi.ai) ## Support * **AnyAPI Dashboard**: [dash.anyapi.ai](https://dash.anyapi.ai) * **AnyAPI Discord**: [Discord community](https://discord.gg/rJgyzGynHw) * **AnyAPI Email**: [support@anyapi.ai](mailto:support@anyapi.ai) * **OpenClaw Issues**: [github.com/openclaw/openclaw/issues](https://github.com/openclaw/openclaw/issues)