API reference
List models
GET /v1/models lists the model ids your key can use. GET /v1/models/{id} returns one.
https://gateway.kalsenti.ai/v1/modelsOn this page
List models
Lists every model you can call right now, sorted by id. The Python and JavaScript snippets use an OpenAI SDK client set up as in the quickstart.
curl https://gateway.kalsenti.ai/v1/models \
-H "Authorization: Bearer $KALSENTI_API_KEY"{
"object": "list",
"data": [
{
"id": "deepseek-v4.1-flash-1017-raw",
"object": "model",
"created": 1791936000,
"owned_by": "kalsenti"
}
]
}idstring- What you put in a request’s
modelfield. createdinteger- When the model was added, in Unix seconds.
owned_bystring- Always
kalsenti.
Prices, context windows, max output and hosting labels aren’t part of this list: they’re on the Models page. Models that need ID verification are listed too. See Models.
Get one model
curl https://gateway.kalsenti.ai/v1/models/deepseek-v4.1-flash-1017-raw \
-H "Authorization: Bearer $KALSENTI_API_KEY"Returns one model in the same shape as an entry in the list. A model that doesn’t exist or can’t be called right now gets a 404 with the code model_not_found. Percent-encode an id if it contains a /.
With the Anthropic SDK
Requests that carry an anthropic-version header, as the Anthropic SDK’s always do, get both endpoints in the Anthropic format instead, errors included. This shape also gives each model’s limits:
{
"data": [
{
"type": "model",
"id": "deepseek-v4.1-flash-1017-raw",
"display_name": "deepseek-v4.1-flash-1017-raw",
"created_at": "2026-10-14T00:00:00.000Z",
"max_input_tokens": 1000000,
"max_tokens": 128000,
"capabilities": null,
"deprecated_at": null
}
],
"has_more": false,
"first_id": "deepseek-v4.1-flash-1017-raw",
"last_id": "deepseek-v4.1-flash-1017-raw"
}max_input_tokensis the context window, ornullif it isn’t recorded.max_tokensis the longest reply the model can write.- Everything comes on one page, so
limit,before_idandafter_idare ignored andhas_moreis alwaysfalse.