Models

GET /models lists what your account can call. It needs no credential, which makes it useful as a connectivity check as well as a catalogue.

curl https://api.bayle.ai/models

{
  "object": "list",
  "data": [
    { "id": "some-model-id", "object": "model", "type": "text" }
  ]
}

For a browsable catalog with pricing, context length, and capabilities, see the Models page.

Model identifiers are opaque

Pass an id back verbatim as the model field of a generation request. Don’t parse it, normalize its case, or infer capability from its name: identifiers change, and the parts that look meaningful aren’t a contract. If you need to know what a model does, read its type.

Pinning a specific id in your configuration is fine, and usually wise for reproducibility. Just re-check it against this endpoint when something stops working, rather than assuming the id is permanent.

Filtering by class

Pass ?type= to narrow the list.

curl "https://api.bayle.ai/models?type=text"

Value	Models for
text	Chat completions
code	Code-oriented completions
embedding	Embeddings
image	Image generation
inpaint	Image inpainting
upscale	Image upscaling
video	Video generation
tts	Text to speech
asr	Speech recognition
music	Music generation
all	Everything, the same as omitting the parameter

An unrecognized value is rejected through the standard error envelope.

Note that a model class appearing here does not mean Bayle serves an endpoint for it. The API surface is the eight operations in the reference; GET /models reports what models exist, not what you can call.

Picking a model in code

models = client.models.list()
text_models = [m.id for m in models.data if getattr(m, "type", None) == "text"]

Or filter server-side, which is cheaper:

import httpx

data = httpx.get("https://api.bayle.ai/models", params={"type": "text"}).json()["data"]