Skip to main content

​
HTTP API: Get Fine-Tune Status

You can monitor the status and progress of a specific fine-tune job via the HTTP API. Endpoint
  • GET /v1/fine_tunes/{fine_tune_id}
Path Parameters
  • fine_tune_id – Unique identifier of the fine-tune job you want to monitor.
Headers
  • Authorization – Bearer token used to authenticate the request (for example Bearer $API_KEY).
  • Content-Type – Must be set to application/json.
Example Request 
curl -X GET "https://api.yourprovider.com/v1/fine_tunes/$FINETUNE_ID" \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json"
Response The endpoint returns a JSON object with the following structure:
  • status (string) – Overall request status (for example "success").
  • status_message (string) – Human-readable description of the status.
  • fine_tune (object) – Details about the fine-tune job.
The fine_tune object contains (non-exhaustive):
  • id (string) – Fine-tune job ID.
  • name (string) – Human-readable name for the fine-tune run.
  • status (string) – Current status (for example "pending", "running", "completed", "failed").
  • base_model (string) – Identifier of the base model (for example "Qwen/Qwen3-0.6B").
  • repository_id (string) – ID of the repository the dataset/model belong to.
  • resource (object) – Input dataset reference:
    • path (string) – Path to the dataset inside the repo.
    • version (string) – Dataset version or commit hash.
  • training_params (object) – Training configuration used for this fine-tune, including:
    • question_column (string) – Dataset column used as the prompt/input.
    • answer_column (string) – Dataset column used as the target/label.
    • batch_size (integer) – Batch size.
    • epochs (integer) – Number of epochs.
    • learning_rate (number) – Learning rate.
    • seq_length (integer) – Maximum sequence length.
    • use_lora (boolean) – Whether LoRA fine-tuning is enabled.
  • gpu_count (integer) – Number of GPUs used.
  • gpu_model (string) – GPU type (for example "A10G").
  • total_token_count (integer) – Total number of tokens processed so far.
  • created_by (object) – User who created the fine-tune (ID, username, name, etc.).
  • source_model (object) – Metadata about the source/base model (ID, name, slug, capabilities, pricing, etc.).
  • inserted_at (string, ISO 8601) – Time the job was created.
  • started_at (string, ISO 8601 | null) – Time the job started running.
  • finished_at (string, ISO 8601 | null) – Time the job finished (if completed/failed).
  • updated_at (string, ISO 8601) – Last time the job metadata was updated.
Other fields such as queue_position, rate_per_second, use_lora, and model pricing/infra metadata may also be present and can be used for advanced monitoring or billing.