Primary sourcerohanpaul_ai
View sourceEvent ·Multi-source
Moonshot AI 发布 Kimi K3.1 Pro:Terminal-Bench 得 65.7%,11 月 4 日开放全量权重
Moonshot AI 发布 Kimi K3.1 Pro,在 Terminal-Bench 4.0 上拿到 65.7%,比 Claude Opus 5.5 低 0.7 分,比 GPT-6 Astra 高 7.5 分,单任务平均成本低于两者。为抑制漏洞利用能力,Moonshot 在中期训练和 RL 阶段移除了网络安全数据,内部 xdev-bench-hard 的 pass@3 从 90.5% 降到 81.0%。工具调用幻觉测试中,虚假陈述比例从 K3 的 23.4% 降到 8.3%,但 Terminal-Bench Science 仅得 38.1,落后 GPT-6 Astra 的 68.1。K3.1 Pro 和 2-3 倍速的 Highspeed 版本已上线 Kimi 应用与 API,全量权重将在 11 月 4 日放出。
Current read
月之暗面更新了 Kimi K3.1 Pro,跑分紧咬 Claude Opus 5.5,成本还更低,11 月 4 日放出全量权重可以自己复测。
11 sources82° AI heatLast updated 2026/10/23 19:21:53
Evidence
11 sourcesRelated source 1Decoder
View sourceRelated source 2Simon Willison’s Weblog
View sourceRelated source 3kimmonismus
View sourceRelated source 4Yuchen Jin
View sourceRelated source 5kimi_moonshot
View sourceRelated source 6TestingCatalog
View sourceRelated source 7berryxia
View sourceRelated source 8Teortaxes
View sourceRelated source 9Artificial Analysis
View sourceRelated source 10Bindu Reddy
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.