Skip to main content
Platform Models are the models already available on BenchGen: published models and endpoints that are ready to evaluate without any extra setup. This is the default source on the Evaluate screen and the quickest way to score a model your workspace already has.
Not the source you need? If you want to pull a model from the Hub, see From a HuggingFace model; if you started a live endpoint yourself, see On a running model.

​
Prerequisites

  • A benchmark or environment to evaluate against, from the Environments Hub or a custom environment you uploaded.
  • At least one model available under Platform Models. See Add a model.

​
Steps

​
1. Open a benchmark

From Eval → Environments, open the benchmark you want to run against. The AI Benchmarks environment list Click Evaluate in the top right, or open the Evaluate tab.

​
2. Select a platform model

The Evaluate tab opens on the Platform Models tab. It lists every model available on the platform, each with a ready badge. Use the search box to filter, then click a model to select it. The Platform Models tab with the model list and Advanced panel Once a model is selected, the Advanced: model environment & parameters panel fills its connection values from that model, and the Run Evaluation button becomes active.

​
3. Set run values (optional)

Expand Advanced: model environment & parameters to review what the benchmark injects:
  • Model (under test) chips are auto-filled from the selected model.
  • You provide these are the fields the benchmark owner exposed to runners.
  • Sampling parameters (Temperature, Max tokens, Top P, Timeout) tune the model connection.
See Environment variables for how these values are declared and injected into the run.
Leave a field blank to fall back to the benchmark defaults. The runner fills in values only; the keys are fixed by the benchmark owner.

​
4. Run and review

Click Run Evaluation. BenchGen creates the run, streams live logs, and computes the score when it finishes. Open the completed run to see the Score Breakdown and the per-question Detailed Results table. See Read results for a full walkthrough of the results page.

​
Next steps

Last modified on July 14, 2026