Meta AI·released 2026-04-08
Scored on 7 benchmarks, leading 1. 71% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on: FrontierMath-2025-02-28-Private
Every score, ranked, colored by evidence status.
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.89accuracy (%) | reproduced· optimized | 2026-04-08 |
| GPQA diamond | 86.4accuracy (%) | reproduced· optimized | 2026-04-08 |
| FrontierMath-2025-02-28-Private | 68.42accuracy (%) | reproduced· optimized | 2026-04-08 |
| SimpleQA Verified | 66.3accuracy (%) | reproduced | 2026-04-08 |
| HLE | 37.56accuracy (%) | unverified | 2026-04-08 |
| FrontierMath-Tier-4-2025-07-01-Private | 24.33accuracy (%) | reproduced | 2026-04-08 |
| CritPt | 11.33accuracy (%) | unverified | 2026-04-08 |