gpt-6IMAGE
Arena 国际象棋测试:Astra 几乎全胜 Opus 5.5
据称 Arena 让 Astra 与 Opus 5.5 对弈,Astra 只输一局;作者还提到 6.1 Sol 登顶对齐指数,并用 Devin Fusion 组合两者。
J A Z I I
@notjazii20 天前收录原推发布于 2026.10.10

J A Z I I @notjazii原推发布于 2026.10.10
openai models are genius arena team ran a chess test between astra and opus 5.5, and astra won all games except one and 6.1 sol topped new alignment index from arena too been using astra and 6.1 sol together via devin fusion and it's the best combo ever made can't wait to see what team cooks next
x.com/notjazii/status/2108906762225209411采集于 2026.10.13 10:10