JSON API Match models across providers (missing `base_model`)

[martinzugnoni]

I can't find a way of matching the canonical model from provider-specific model data in the JSON API.

Looking at the .toml files in the repo it seems that the base_model field is used to link back to the canonical model data
https://github.com/anomalyco/models.dev/blob/dev/providers/openrouter/models/openai/gpt-5.5.toml

base_model = "openai/gpt-5.5"

Points to: https://github.com/anomalyco/models.dev/blob/dev/models/openai/gpt-5.5.toml

The problem is that base_model is not exposed in any of the 3 JSON APIs, and "id" in each provider model document in the API response is provider-specific and not the canonical model ID.

What would be the correct way of matching models across providers in the JSON API?

Thanks!

[Bhavysinghal]

Hi, I'd like to work on this issue!

After looking at the code, I found that base_model is intentionally stripped in mergeBaseModel() inside generate.ts:

const { base_model: _baseModel, base_model_omit: omit, ...overrides } = model;

There's also an existing test that explicitly asserts base_model should not appear in the JSON output:

expect(catalog.providers.factored?.models.model).not.toHaveProperty(
  "base_model",
);

(packages/core/test/generate.test.ts, line 91)

Before proceeding, could you clarify the intended approach?

• Should base_model be exposed in the JSON output and the existing test updated to reflect this?
• Or is there a different approach you have in mind (e.g. a separate lookup field, or a different API endpoint)?

Thanks!