OpenAI's 6.2-Astra runs 8 hours on AtCoder contest, produces bizarre compressed solution
i_dg23 · x · 2026-11-19
Japanese developer @wataorz let OpenAI's 6.2-Astra run for 8 hours on an AtCoder heuristic contest, producing what he calls an incomprehensibly over-compressed solution — "6.2-Astra is a monster."
The attached leaderboard shows the OpenAI account sitting at rank 7 with 6.03 billion points, alongside top human heuristic competitors on the long-running contest.
More from Models
- Contra Labs launches Frontier Design Benchmark judged by real designers — briannekimmel · 2026-11-19
- Clairic open-weights extraction models launch in 27B and 9B sizes — teortaxesTex · 2026-11-19
- New Astra mocked for misreading "all numbers in the paper" as excluding page numbers — analisereal · 2026-11-19
- Gemini 4 pricing criticized: $7.50-$20 per M output tokens, no cheap or flagship tier — haider1 · 2026-11-19
- Gensyn launches Open8B, the first fully auditable LLM you can verify on a gaming GPU — benfielding · 2026-11-19
- Stealth model 'Tidepool Alpha Free' appears on OpenCode Zen, with native image gen — gnukeith · 2026-11-19