Skip to main content
The Training Jobs API manages the full lifecycle of a fine-tuning job. All endpoints use the /v1/training-jobs route family and require an API key in the X-API-Key header. All training operations use the same API authentication as inference. API base URL: https://api.fastino.ai
Send the key on every training request:
Use GET /v1/base-models?supports_training=true immediately before creating a job. It is the live source of truth for trainable model IDs. Before submitting a job, prepare and upload a training dataset and wait for its processed version to become ready.
The legacy /felix/training-jobs route family is retired. Use /v1/training-jobs for new and migrated integrations.

​
Training workflow

Choose a base model

List models currently available for training.

Upload a dataset

Reserve an upload URL, upload JSONL, and process it.

Create a job

Submit datasets and training configuration.

Monitor and manage

Poll status, inspect logs and billing, then work with artifacts.

​
Status lifecycle

Training-job responses include:
  • status: persisted job status.
  • normalized_status: stable status intended for client polling.
  • is_terminal_status: whether polling can stop.
  • error_message: failure information when available.
  • is_deployable: whether the completed artifact passes deployment validation.
  • deployability_reason: why the artifact is not deployable.
Prefer normalized_status and is_terminal_status in polling clients.

​
Endpoints