Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mulerouter.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

Every MuleRouter endpoint sits under a single host: 
https://api.mulerouter.ai
The path is what tells MuleRouter which provider and which model you want. See the API Reference for the exact URL of every endpoint.

​
Industry aliases

For the few endpoints that have become de-facto industry standards, we expose short aliases at the root so existing SDKs work with a single base_url swap.
AliasSource
/v1/chat/completions/vendors/openai/v1/chat/completions
/v1/models/vendors/openai/v1/models

​
Synchronous vs asynchronous endpoints

  • Synchronous / streaming. Chat completions and similar fast endpoints respond within a single HTTP request — switch the base URL and you’re done.
  • Asynchronous tasks. Image, video, speech, and music generation return a task_id immediately and finish out-of-band. Poll GET .../<task-id> (or use webhooks) to retrieve the result. See Asynchronous tasks for the full pattern.

​
Response headers worth knowing

HeaderWhat it’s for
TraceparentA W3C-format trace ID for this request. Include it when filing bug reports so we can find the exact run in our logs.

​
Migrating from a direct provider integration

In most cases, all you need to change is the base_url. For example, moving from OpenAI direct to MuleRouter:
python
- client = OpenAI(api_key=OPENAI_KEY)
+ client = OpenAI(
+   api_key=MULEROUTER_KEY,
+   base_url="https://api.mulerouter.ai/v1",
+ )
For providers that have their own request shape (Google, Kling, Wan, etc.), use the matching /vendors/<provider>/... path. The per-endpoint reference pages list the exact request and response schemas.