SatBench: 63% of AI Benchmarks Cited by Labs Are Saturated

Ethan Mollick's SatBench project finds 63% of the 173 benchmarks cited by major AI labs since 2023 are saturated, and private held-out benchmarks degrade as fast as public ones.

2026-11-15 ~ 2026-11-15 · 3 related posts

• SatBench: 63% of lab-cited AI benchmarks already saturated, and solving is speeding up — emollick · 2026-11-15
• Private AI benchmarks start ~11 points harder but saturate just as fast as public ones — emollick · 2026-11-15

1 near-duplicate retellings: emollick