> ## Documentation Index
> Fetch the complete documentation index at: https://docs.antarah.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Models

> Antara's models, their capabilities, and how to read the live catalog.

Call a model by its ID: `a1`, `a2`, `a3`, or `a4`. The catalog below is today's; `GET /v1/models` is always current, with prices.

| ID | Input | Tools | Reasoning efforts | Context window | Max output |
| - | - | - | - | - | - |
| `a1` | text | yes | `none`, `low`, `medium`, `high` | 1,048,576 | 131,072 |
| `a2` | text, image | yes | `none`, `low`, `medium`, `high` | 1,048,576 | 131,072 |
| `a3` | text | yes | — | 262,144 | 65,536 |
| `a4` | text | no | — | 131,072 | 16,384 |

## The live catalog

```bash theme={null}
curl https://api.antarah.dev/v1/models -H "Authorization: Bearer $ANTARA_API_KEY"
```

```json theme={null}
{
  "object": "list",
  "data": [{
    "id": "a1",
    "object": "model",
    "created": 1793750400,
    "owned_by": "antara",
    "display_name": "Antara A1",
    "context_window": 1048576,
    "max_output_tokens": 131072,
    "input_modalities": ["text"],
    "output_modalities": ["text"],
    "supports_tools": true,
    "supports_reasoning": true,
    "reasoning_efforts": ["none", "low", "medium", "high"],
    "pricing": { "input_per_mtok": "1.25", "cached_input_per_mtok": "0.20", "output_per_mtok": "2.50" },
    "deprecated": false,
    "sunset_at": null,
    "access": { "available": true, "locked": false, "required_checks": [], "missing_requirements": [] }
  }]
}
```

| Field | Meaning |
| - | - |
| `input_modalities` | `image` means the model accepts [images](https://docs.antarah.dev/guides/images.md). Sending one to a text-only model gets `400 unsupported_modality` |
| `supports_tools` | Whether `tools` is accepted. If not, sending tools gets `400 unsupported_parameter` |
| `reasoning_efforts` | The values `reasoning_effort` accepts for this model. See [Reasoning](https://docs.antarah.dev/guides/reasoning.md) |
| `pricing` | USD per million tokens, as decimal strings. Cached input is input the model runtime served from its prompt cache |
| `deprecated`, `sunset_at` | A deprecated model still works until `sunset_at`, then answers `410 model_retired`. See [Versioning](https://docs.antarah.dev/guides/versioning.md) |
| `access` | Whether your organization can call the model. A locked model answers `403 model_locked` |

Responses always name the public model (`"model": "a1"`), never the runtime behind it.

> [!NOTE]
>
> `/v1/models` is cached for 60 seconds per project and sends an `ETag`, so polling it is cheap.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.