Event ·Multi-source

Cognition 发布 SWE-2 模型，在编码基准测试中表现强劲

Cognition 的 SWE-2 模型（基于 Kimi K3 后训练）在 FrontierCode 基准测试中取得 50.00% 的分数，与 Fable 5.1 相当，但成本降低了 64%。该模型在多项评估中与前沿模型表现相当，成本可降低 70%。

Current read

朋友发了 Cognition 新的 SWE-2 模型，这个模型在编码方面很厉害，和 Fable 5.1 差不多，但成本更低。

9 sources58° AI heatLast updated 2026/9/10 16:44:04

Evidence

9 sources

Primary sourceThe Rundown AI

2026/9/10 16:44:04View source

Related source 1Cognition

2026/9/10 15:21:01View source

Related source 2elvis

2026/9/10 16:00:30View source

Related source 3arXiv cs.AI

2026/9/8 17:25:47View source

Related source 4Blognone

2026/9/10 15:30:28View source

Related source 5IT之家

2026/9/11 10:03:41View source

Related source 6Fireworks AI

2026/9/8 18:21:50View source

Related source 7OpenRouter

2026/9/9 14:57:26View source

Related source 8Windsurf

2026/9/10 16:52:47View source

Conflict 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.

Open the story