Fixing "Unknown model" (model_not_found)

This error means the model field held an id the API does not serve. The key and balance were fine; the name was not.

What the error says

On /v1/chat/completions it is HTTP 400:

{
  "error": {
    "message": "Unknown model \"gpt-4o\". Fetch GET /v1/models for the catalogue.",
    "type": "invalid_request_error",
    "code": "model_not_found",
    "param": null
  }
}

On /v1/messages the same case is a 404 not_found_error, which is what Anthropic clients expect. If the field is missing entirely, you get 400 Field "model" is required. instead.

Ids this API accepts

• outlaw-1
• glm-5.3-outlaw (text only)
• glm-5.3-flash-outlaw
• mimo-v2.6-flash-outlaw
• qwen3.8-27b-outlaw

Ids are matched exactly, so case and spelling matter. Only the models in this line are served; naming some other provider's model id, even one you know exists elsewhere, is refused with this error. The live list, with context sizes and prices, is at GET https://wildwestapi.com/v1/models and on the models page.

Common causes

• A leftover default. SillyTavern and many SDK examples default to an OpenAI model name. If you switched the source to Custom but never set the model, the old name goes out.
• Typos and display names. "Outlaw 1" is a display name; the id is outlaw-1.
• An outdated preset that stored a model id which is no longer offered.

Step by step fix

1. In SillyTavern, open API Connections with Custom (OpenAI-compatible) selected and press Connect. The model list loads from /v1/models.
2. Pick a model from that list, or type the id exactly into the model field.
3. In code, set model="outlaw-1" (or another listed id) on the call itself.
4. Send a test message.

For reference, the working SillyTavern setup is: API set to Chat Completion, source set to Custom (OpenAI-compatible), endpoint https://wildwestapi.com/v1, your sk-ww- key in the API key field, and a model id such as outlaw-1. The full walkthrough is on the SillyTavern setup page.

Telling it apart from similar errors

A 404 with No route for POST ... is a wrong URL, not a wrong model; see 404 No route. A 400 that names a model and mentions images is a capability problem, covered in model does not accept images.

FAQ

Related