API reference · API
Retrieve one model
GET/v1/models/{id}No key
The entry GET /v1/models lists, for a chat model, an autopilot id or an embedding model. The id carries its vendor and a slash (openai/gpt-4o), sent as two path segments or as one encoded segment, which is what the OpenAI clients send. An id that is not in the catalog is a JSON 404 with code model_not_found and up to three ids you may have meant in the message.
Parameters
| Field | Type | What it is |
|---|---|---|
| id (path)required | string | A catalog id such as openai/gpt-4o. |
Answer 200
The model.
Headers: x-antimatter-request-id, x-request-id
| Field | Type | What it is |
|---|---|---|
| id | string | Namespaced as vendor/model. |
| object | "model" | |
| created | integer | Unix seconds when the upstream first listed it, where it says. |
| owned_by | string | The serving adapter: openai, anthropic, openrouter, or antimatter. |
| name | string | |
| context_length | integer or null | |
| max_output_tokens | integer | The most tokens one answer may generate on this model, where the upstream publishes it. max_tokens is clamped here. Absent when unpublished. |
| pricing | Pricing | |
| pricing. | number or null | USD per million prompt tokens. Null when the upstream publishes no price. |
| pricing. | number or null | USD per million completion tokens. Null when the upstream publishes no price. |
| pricing. | number or null | USD per million prompt tokens the vendor served from its own cache. Absent or null when the vendor publishes none, and such tokens bill at the prompt rate. |
| pricing. | number or null | USD per million prompt tokens written into the vendor's cache. Absent or null when the vendor publishes none, and such tokens bill at the prompt rate. |
| pricing_unknown | boolean | True when no price could be established for this model. It is refused on house money (503 pricing_unavailable) and must never be read as free. Absent on the autopilot entries. |
| modality | string | One of "text", "multimodal". |
| output_modalities | array of string | What the model produces, where the upstream says: text, image, speech, transcription or embeddings. Image, speech and transcription models are served by /v1/images/generations and /v1/audio/*, embedding models by /v1/embeddings and listed at GET /v1/embeddings/models. |
| input_modalities | array of string | What the model reads, where the upstream says: text, image, audio, file or video. Absent when the upstream does not say. |
| holders_only_until | integer | Unix seconds when a model still in its $ANTI holders' window opens to everyone. Until then it serves holders only (the account that minted the key holds the lowest holder tier) and anyone else is refused with 403 holder_only; autopilot never routes a non holder to it. Absent on every other model, and on a deployment without holder tiers. |
Errors
- 404OpenAI shaped error.
Every error answers in one shape, with its type, code and the field it names. The whole list is in the docs' Errors section.