GPT-6.2 Astra slumps to 14th on SimpleBench at 79.2%, behind its own predecessor
scaling01 · x · 2026-11-19
The SimpleBench leaderboard (AVG@5) shows OpenAI's newest models struggling:
- GPT-6.2 Astra Pro lands 14th at 79.2%, with GPT-6.2 Astra at 15th (77.3%) — both below the 83.7% human baseline and behind previous versions like GPT-6 Astra Pro (86.5%, 6th)
- Top spot goes to Anthropic's Claude Fable 5.2 at 94.9%, near the 95.4% top human score; Google's Gemini 4 Carbon and Argon follow at 92.6% and 91.7%
The poster asks "what is OpenAI doing" — a notable regression for the newest flagship models.
Related event: GPT-6.2 Astra regresses sharply on SimpleBench(2 posts)→
More from Models
- Three Google flagship models in eight days fuel speculation about hidden RSI work — haider1 · 2026-11-19
- Mystery Model 'Gestalt' Tops New Frontier Design Benchmark by 214 Points — soleio · 2026-11-19
- mrfakename releases abliterated GLM-5.3-Flash and Qwen3.8-Flash-Next — mrfakename0 · 2026-11-19
- Physician: AI competition now measured in weeks, GPT-6 versions 7 days apart — ramos_casals · 2026-11-19
- micro1's flow qc model screens training data for benchmark leaks before release — Exp_Mark · 2026-11-19
- Leaker predicts new DeepSeek paper this week, scheduled for November 11 — teortaxesTex · 2026-11-19