Models

GET /v1/models lists the active models in the catalogue. It requires a user API key.

Request

curl https://<your-domain>/v1/models \
  -H "Authorization: Bearer $EASYAI_API_KEY"

Response

{
  "object": "list",
  "data": [
    {
      "modelKey": "lfm3",
      "revision": "<revision>",
      "architecture": "<architecture>",
      "quantization": "<quantization>",
      "status": "active",
      "createdAt": "<ISO timestamp>"
    }
  ]
}

Only rows with status: "active" are returned, ordered by model key and revision. createdAt is an ISO timestamp. The exact revision, architecture, and quantization values are database values, so read them from the response rather than hard-coding them in a client.

Live and planned models

• LFM3-1.7B is live with the model key lfm3.
• Qwen3 models are planned. They are not live and are not returned by this endpoint.

The catalogue describes models the control plane knows about. It does not promise that a worker is online at this instant. A chat request can still fail with 503 NO_CAPACITY when no eligible worker has the model loaded.

Using a model key

Pass a key from this list in the chat request:

{
  "model": "lfm3",
  "messages": [{ "role": "user", "content": "hello" }]
}

Unknown or inactive model keys fail with 400 VALIDATION_FAILED. See Chat completions and Errors.