Skip to main content

Model Evaluations

Evaluation runs and benchmark results for deployed models. These endpoints let you create evaluation jobs, retrieve results, manage custom evaluation definitions, and track evaluation progress.

Standard Evaluations


POST

Create model evaluation

Starts a new benchmark evaluation run for a deployed fine-tuned model.

GET

Get evaluation data

Returns evaluation results and benchmark scores for a fine-tuned model identified by its numeric model ID.

GET

Get evaluation data by model name

Returns evaluation results and benchmark scores for a fine-tuned model identified by its model name.

Custom Evaluations


POST

Create custom evaluation

Creates a new custom evaluation definition with a natural-language scoring prompt that defines how outputs are compared.

GET

Get custom evaluation

Returns all custom evaluation definitions created by the authenticated user, including the evaluation name and scoring prompt for each.

GET

Get custom evaluation results

Returns all recorded results for a custom evaluation run, including per-run win/draw/loss counts and aggregate improvement percentage.

POST

Start custom evaluation

Starts a custom evaluation run that compares two models head-to-head using the specified evaluation criteria.