Bring Your Own LLM

By default, a workspace’s playground/chat uses the platform’s default model. Set a per-workspace OpenRouter API key to route that workspace’s chat through your own OpenRouter account, and (optionally) choose the model. This is configured on the workspace endpoints (POST /api/v1/mcp/workspaces and PATCH /api/v1/mcp/workspaces/{id}). The environment (sandbox/production) is always derived from the API key — sk_test_* keys operate on sandbox, sk_live_* on production.

Looking to disable server-side conversation storage and supply prior turns yourself? See the Stateless Chat guide.

Set the key on create

curl -X POST https://api.routemcp.io/api/v1/mcp/workspaces \
  -H "Authorization: Bearer sk_test_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Support Bot",
    "integrations": ["github", "gmail"],
    "openRouterApiKey": "sk-or-v1-xxxxxxxxxxxxxxxxxxxxxxxx",
    "model": "anthropic/claude-sonnet-4-6"
  }'

Update the key or model later

curl -X PATCH https://api.routemcp.io/api/v1/mcp/workspaces/{id} \
  -H "Authorization: Bearer sk_test_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemini-3.5-flash"
  }'

The dashboard exposes the same controls in the workspace’s AI Config card.

Field reference

Field	Type	Notes
`openRouterApiKey`	string	Write-only. Validated and encrypted at rest; never returned in any response. Setting it replaces the workspace’s existing key. Optional.
`model`	string	OpenRouter model id in `vendor/model` format (e.g. `anthropic/claude-sonnet-4-6`, `google/gemini-3.5-flash`). Optional.

openRouterApiKey is input-only. It is never echoed in create/get/update/list responses. To inspect the current configuration, read the llm status object (below). To remove a key, call DELETE /api/v1/mcp/workspaces/{id}/llm-key.

Platform fallback

When no openRouterApiKey is configured on a workspace, chat falls back to the platform’s default model — you don’t have to set a key to use chat at all.

The `llm` status object

Create, get, and update responses (and each list item) include an llm object describing the workspace’s BYO LLM state. The key itself is never present here.

{
  "data": {
    "id": "0191e6a2-1c4d-7a3b-9f10-2b3c4d5e6f70",
    "name": "Support Bot",
    "statelessMode": false,
    "llm": {
      "configured": true,
      "model": "anthropic/claude-sonnet-4-6",
      "verificationStatus": "verified",
      "verifiedAt": "2026-05-29T10:05:00.000Z",
      "updatedAt": "2026-05-29T10:05:00.000Z"
    }
  }
}

Field	Type	Meaning
`configured`	boolean	Whether a per-workspace OpenRouter key is set.
`model`	string \| null	The configured OpenRouter model id, or `null` when none is set.
`verificationStatus`	`"verified"` \| `"invalid"` \| `"unverified"` \| null	`verified` — the key works; `invalid` — the key was rejected; `unverified` — not yet checked; `null` — no key configured.
`verifiedAt`	string \| null	When the key was last successfully verified, if ever.
`updatedAt`	string \| null	When the LLM key/model was last changed.

RouteMCP currently supports OpenRouter only for bring-your-own LLM keys. Support for additional providers is in progress and coming soon.

Getting Started

Guides

References

Changelog

Set the key on create

Update the key or model later

Field reference

Platform fallback

The `llm` status object

​Set the key on create

​Update the key or model later

​Field reference

​Platform fallback

​The llm status object

Set the key on create

Update the key or model later

Field reference

Platform fallback

The `llm` status object