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

# Discover PIN Training Capabilities and Quotas

> Return the catalog of supported models, training methods, LoRA options, export formats, quota, balance, and pricing for the current user.

`GET /v1/capabilities` returns a discovery snapshot for the authenticated user: allowed base models, training methods, LoRA methods, export formats, target modules, quota usage, wallet balance, and pricing hints including whether auto-intervention is enabled.

## Method + Path

```http theme={null}
GET /v1/capabilities
```

## Authentication

```http theme={null}
Authorization: Bearer <jwt-or-api-key>
```

## Response

Standard PIN envelope. `data` fields:

<ResponseField name="models" type="array of strings">
  Base model allowlist from server config.
</ResponseField>

<ResponseField name="methods" type="array of strings">
  Supported job types (e.g. `sft`, `opd`, `grpo`, `dpo`, `pretrain`, `distill`, `qat`, …).
</ResponseField>

<ResponseField name="lora_methods" type="array of strings">
  LoRA method allowlist (`lora`, `qlora`, `dora`, `none`).
</ResponseField>

<ResponseField name="export_formats" type="array of strings">
  Export formats (`gguf`, `mlx`, `mnn`, `qnn`).
</ResponseField>

<ResponseField name="target_modules" type="array of strings">
  Allowed LoRA target module names.
</ResponseField>

<ResponseField name="quota" type="object">
  `total` and `used` token quota for the current user.
</ResponseField>

<ResponseField name="balance" type="number">
  Wallet balance.
</ResponseField>

<ResponseField name="pricing" type="object">
  `gpu_hour_usd`, `token_per_m_usd`, and `auto_intervention` (`enabled`, `conditions`).
</ResponseField>

## Example

```bash theme={null}
curl -s http://localhost:8001/v1/capabilities \
  -H "Authorization: Bearer $PIN_API_TOKEN"
```

```json theme={null}
{
  "code": 0,
  "data": {
    "models": ["aria/AFM-2.6B", "LiquidAI/LFM2-2.6B"],
    "methods": ["sft", "opd", "grpo", "dpo", "kto", "orpo", "simpo", "ppo", "pretrain", "distill", "qat"],
    "lora_methods": ["lora", "qlora", "dora", "none"],
    "export_formats": ["gguf", "mlx", "mnn", "qnn"],
    "quota": { "total": 1000000000, "used": 0 },
    "balance": 0,
    "pricing": {
      "gpu_hour_usd": 1.2,
      "token_per_m_usd": 0.1,
      "auto_intervention": {
        "enabled": true,
        "conditions": ["spike", "drop", "kl_explode", "len_surge", "repeat_up"]
      }
    }
  },
  "message": ""
}
```


## Related topics

- [Aria Compute PIN API: Post-Training & Evaluation](/api-reference/pin/introduction.md)
- [Aria Compute platform changelog](/resources/changelog.md)
- [Aria Compute ROUTER: OpenAI-compatible inference gateway](/api-reference/router/introduction.md)
- [Restart PIN Training Agent Process](/api-reference/pin/agents/restart.md)
- [Create Training Job with PIN API](/api-reference/pin/jobs/create.md)