Models API

/ API docs

GET https://api.opencheese.dev/v1/models lists the models your API key can use right now, in the OpenAI list shape. What it contains depends on your plan, the key's model restrictions and the models you switched off on the Prices page, so read it instead of copying model IDs from examples.

Request

bash

curl https://api.opencheese.dev/v1/models \
  -H "Authorization: Bearer $OPENCHEESE_API_KEY"

x-api-key works too. An invalid key answers 401 with a JSON error.

Response

A text model (the values are examples):

json

{
  "object": "list",
  "data": [
    {
      "id": "MODEL_ID",
      "object": "model",
      "created": 0,
      "owned_by": "opencheese",
      "displayName": "Example Model",
      "name": "Example Model",
      "capabilities": { "streaming": true, "tools": true },
      "pricing": { "inputRate": 0.5, "outputRate": 2, "cacheReadRate": 0.05, "cacheWriteRate": 0.6 },
      "modalities": { "input": ["text", "image"], "output": ["text"] },
      "reasoning": true,
      "reasoningEfforts": { "max": "max", "medium": "medium", "low": "low" },
      "supported_reasoning_levels": ["max", "medium", "low"],
      "efforts": ["max", "medium", "low"],
      "default_reasoning_level": "max",
      "context_window": 1000000,
      "contextWindow": 1000000,
      "max_output_tokens": 131000,
      "maxTokens": 131000,
      "maxOutputTokens": 131000
    }
  ]
}

Fields

Field	Meaning
`id`	The ID to send as `model`.
`displayName`, `name`	The model's name as the dashboard shows it.
`capabilities`	Whether the model streams and calls tools.
`pricing`	Your rates in credits (USD) per million tokens: input, output, cached input read and cache write. Free models show zeros.
`modalities`	What the model reads (`input`) and produces (`output`): `text`, `image`, `video`, `audio`, `pdf`.
`supported_reasoning_levels`	The effort levels you may send, strongest first. `efforts` and the keys of `reasoningEfforts` repeat them for clients that read those names.
`default_reasoning_level`	The first, strongest level of that list.
`context_window`	The most tokens the model reads in one request, prompt included. `contextWindow` is the same number.
`max_output_tokens`	The longest answer the model writes. `maxTokens` and `maxOutputTokens` are the same number.

A field is left out when it does not apply: a model with no selectable effort has no level fields, and a model without a known limit has no limit fields. Send the effort only with a level from supported_reasoning_levels; Chat Completions, Responses and Messages show where it goes.

Image models

Models whose modalities.output is ["image"] generate pictures. They carry only the basic fields above; their parameters (sizes, aspect ratios, quality, how many images a request may return) come from GET /v1/images/capabilities, described in the Image API.

Filter the list

All text models with jq:

bash

curl -s https://api.opencheese.dev/v1/models \
  -H "Authorization: Bearer $OPENCHEESE_API_KEY" \
  | jq -r '.data[] | select(.modalities.output | index("text")) | .id'

Free models:

bash

curl -s https://api.opencheese.dev/v1/models \
  -H "Authorization: Bearer $OPENCHEESE_API_KEY" \
  | jq -r '.data[] | select(.pricing.inputRate == 0 and .pricing.outputRate == 0) | .id'

Free models still need a small minimum balance on your account; see the FAQ.

With an SDK

python

from openai import OpenAI

client = OpenAI(base_url="https://api.opencheese.dev/v1", api_key="YOUR_OPENCHEESE_API_KEY")
for model in client.models.list():
    print(model.id)

There is no GET /v1/models/{id}: read the model from the list.

The cheese has holes. Your infrastructure should not.