RL

Returns metadata for a checkpoint: type, base model, LoRA rank, step, and owning session.

get/v1/rl/checkpoints/{id}

Request

  • Base URL
    https://api.together.ai
  • URL
    https://api.together.ai/v1/rl/checkpoints/{id}
  • Auth
    HTTP bearer

Path parameters

idstring required

ID of the checkpoint

Response

Checkpoint metadata

idstring required

Unique identifier for the checkpoint

session_idstring required

Training session that produced the checkpoint

base_modelstring required

Base model the checkpoint was trained from

type'CHECKPOINT_TYPE_TRAINING' | 'CHECKPOINT_TYPE_INFERENCE' required

Whether a checkpoint is saved for training resume or inference download.

lora_rankinteger

LoRA rank of the session that produced this checkpoint. Absent for full-weight sessions and for checkpoints saved before this field was recorded.

created_atstring date-time required

Timestamp when the checkpoint was created

Example response

{
  "id": "123e4567-e89b-12d3-a456-426614174000",
  "session_id": "123e4567-e89b-12d3-a456-426614174000",
  "base_model": "Qwen/Qwen3.5-4B",
  "lora_rank": 32,
  "created_at": "2026-01-02T00:00:00Z",
  "inference_registration": {
    "model": {
      "id": "ml_abc123",
      "revision_id": "rv_def456"
    },
    "adapter": {
      "id": "ml_abc123",
      "revision_id": "rv_def456"
    }
  },
  "training_registration": {
    "id": "ml_abc123",
    "revision_id": "rv_def456"
  }
}

Changes