You can now pause and resume benchmarks

Not every benchmark needs to run forever. Some questions have a natural measurement window – you run for a few days, get your answer, and stop. Until now, the only options were delete the benchmark (losing your history) or leave it running and consuming usage you did not need.

Pause stops the schedule without touching anything else. Your run history, leaderboard, and analytics are all intact. Resume restarts on the same cadence from the next interval.

The pattern this unlocks

A useful pattern for active development:

1. Create a benchmark with a daily schedule
2. Run it for several days, or fire a whole batch of tests using the Run Now button, to establish a baseline
3. Pause to stop consuming usage
4. Resume when you want to measure the effect of a change

This is benchmarking as a test bed rather than a continuous monitor. Same reproducible setup, same rubric, same models, but on your terms, not a fixed clock. If you are actively iterating on a prompt, a store, or a system prompt, this is the faster feedback loop.

Pause and resume are in the Config tab of any benchmark drawer. Full details in the Benchmark Guide.