Sentry ran 10k+ AI evals in 90 days, over half in the past month alone

zeeg · x · 2026-11-20

Sentry engineer @rbro112 shared that the company's internal AI evaluation pipeline surpassed 10,000 eval runs over the last 90 days, with more than 50% of those happening in just the past month — a clear acceleration.

A dashboard screenshot shows the details: 10,004 runs, an 81.7% pass rate, and roughly 20.7B tokens consumed. Daily run volume has been steadily active since late August, peaking above 400 runs per day.

Co-founder zeeg amplified the stat. It's a concrete data point on how evals are becoming standard infrastructure for engineering teams adopting AI coding tools at scale.

Original post →

More from coding & agent

coding & agent channel →