CritPt v2 launches: GPT-6 Astra tops research-level physics at 83%, 54 points above GPT-5
ArtificialAnlys · x · 2026-10-23
Artificial Analysis launched CritPt v2 with the CritPt authors: all 70 research-level physics challenges were reviewed for errors and ambiguity, and models now solve them agentically with code execution in an offline sandbox.
Key results:
- GPT-6 Astra (max) leads at 83%, followed by GPT-6.1 Sol (max) at 82%, Claude Opus 5.5 (max with fallback) at 78%, GPT-5.6 Sol (max) at 75%
- GPT-5 (high), released August 2025, scores just 29%—54 points behind the frontier
- The Intelligence Index will adopt CritPt v2 in late October
- v2 scores aren't directly comparable to v1; most of the gain comes from revised challenges, with tool access adding a smaller boost
Related event: CritPt v2 Physics Benchmark Released; GPT-6 Astra Tops at 83%(4 posts)→
More from Models
- Dev jokes his true calling is telling everyone to try the latest DeepSeek — intellectronica · 2026-10-23
- SubQ Snap 1.1 ties for first on Artificial Analysis Decision Index with 384K context — ArtificialAnlys · 2026-10-23
- PolyAI unveils Dialog-RSN-1: audio-native voice model answering in under 300ms — matthen2 · 2026-10-23
- Konrad Kording: little evidence that chatting with AI personalizes it — KordingLab · 2026-10-23
- Meta paper ditches tokenizers: distilled 1B byte models outperform token-based peers — mtizard · 2026-10-23
- Terminal-RSI lifts Qwen3.8-27B from 51.7% to 86.52% on Terminal-Bench, plus 70 tok/s on M5 Max via DFlash 2 — udmrzn · 2026-10-23