API reference
Fine-tuning
16 operations. Each page has the request, the response, the errors and examples in curl, TypeScript, Python and the CLI.
- GETModels that can be fine-tuned, how, on which hardware and at what price
/v1/fine-tunes/base-models - GETPrice estimate for fine-tuning on a validated dataset
/v1/fine-tunes/estimate - GETList fine-tuning jobs, newest first
/v1/fine-tunes - POSTCreate a fine-tuning job (reserves its estimated cost; 402 when credit is short)
/v1/fine-tunes - GETGet a fine-tuning job
/v1/fine-tunes/{fine_tune_id} - POSTCancel a job (charged only for the tokens it trained; idempotent)
/v1/fine-tunes/{fine_tune_id}/cancel - GETA job's events, oldest first
/v1/fine-tunes/{fine_tune_id}/events - GETA job's training metrics, one point per step in step order (chart-ready)
/v1/fine-tunes/{fine_tune_id}/metrics - GETA job's checkpoints, oldest first
/v1/fine-tunes/{fine_tune_id}/checkpoints - GETDownload the adapter or the merged weights of a checkpoint as a tar archive
/v1/fine-tunes/{fine_tune_id}/download - POSTA short-lived link to download an artefact tar without credentials
/v1/fine-tunes/{fine_tune_id}/download-link - GETThe project's fine-tuned models (results of succeeded jobs), newest first
/v1/fine-tuned-models - GETGet a fine-tuned model of the project
/v1/fine-tuned-models/{fine_tuned_model_id} - DELETEDelete a fine-tuned model (refused with 409 while it is served)
/v1/fine-tuned-models/{fine_tuned_model_id} - POSTServe a LoRA fine-tune per token on the shared pool of its base model (idempotent)
/v1/fine-tuned-models/{fine_tuned_model_id}/enable - POSTStop serving a fine-tune per token (idempotent)
/v1/fine-tuned-models/{fine_tuned_model_id}/disable