GPT-6 Astra Reportedly Beats Physics Benchmark by 80%, Proving Experts Wrong
haider1 · x · 2026-10-16
Per the shared screenshots, GPT-6 Astra reportedly flagged expert results as wrong on a physics benchmark (+80%), and a team verified the model was right — new score: 98% (arxiv 2610.16733). Noam Brown compares this to GPT-5.5 finding errors in FrontierMath, where Epoch AI's review flagged fatal errors in about a third of problems. The poster argues math was just the beginning and physics may follow the same trajectory — a possible path toward recursive self-improvement.
More from Models
- Skeptic doubts rumored OpenAI model, citing inability to serve Astra reliably — CtrlAltDwayne · 2026-10-25
- AsciiBench: blind voting arena tests which AI draws better ASCII art across 13 models — JasonBotterill · 2026-10-25
- Satirical ARC-AGI-3 leaderboard shows GPT-6 near 90% score at $10K+ per task — haider1 · 2026-10-25
- Rumor: Anthropic to drop Opus 5.5 two weeks after calling for frontier pacing — StewartalsopIII · 2026-10-25
- Anthropic appears to be quietly nerfing Opus 5.5 usage limits, users report — CtrlAltDwayne · 2026-10-25
- Codex user reports usage down 31% after reset, calls OpenAI pricing uneconomical — CtrlAltDwayne · 2026-10-25