Models

List models

GET https://api.tokamak.sh/v1/models

curl -sS https://api.tokamak.sh/v1/models \
  -H "Authorization: Bearer $TOKAMAK_API_KEY"

The response contains model entries in data. Copy an entry's id into the model field of your inference request. An id can be an alias the platform publishes for a model (for example a spelling with a [1m] suffix); it serves that model and is recorded under the model's own id. With X-PROVIDER-DATA: true the entry shows the canonical id and the alias in alias. Inactive models are not listed and cannot be called.

An empty catalog returns a successful response with an empty data array.

Choose a model

1. List models from the environment you will use.
2. Select a model that supports your client's API format.
3. Send a small request with an output limit.
4. Add optional features after verifying the basic request.

Model IDs and availability can differ between staging and production. A listed model does not guarantee support for every endpoint or parameter.

See Chat, Responses and Messages for request examples.

Provider information

Use GET /v1/models/providers to list providers represented in the catalog.

For separate provider-model entries instead of the default merged model list, add:

X-PROVIDER-DATA: true

These endpoints describe the deployment's catalog. Actual request success also depends on provider availability and your account's limits and credit.

Troubleshooting

If a model is rejected, check the exact ID, endpoint and supported fields. If the catalog is empty, confirm that you are calling the correct environment.

Continue with Quickstart or Errors and debugging.