SWE ProBrowse 296

SWE Pro

Can the model complete substantial programming work under this benchmark's agent setup?

Version not specifiedExact reported variant
Coding7 ranked models11 reported values1 reportsHigher is better

Top rankings

One row per model, using its best reported score across effort settings.

7
RankModelBest scoreBest reported setting
1Kimi K2.6Moonshot AI58.6%Reported configurationDeepSeek report1 value · 1 reportApr 26, 2026 · source
2GLM-5.1Z.ai58.4%Reported configurationDeepSeek report1 value · 1 reportApr 26, 2026 · source
3GPT-5.4OpenAI57.7%xhigh effortDeepSeek report1 value · 1 reportApr 26, 2026 · source
4Claude Opus 4.6Anthropic57.3%max effortDeepSeek report1 value · 1 reportApr 26, 2026 · source
5DeepSeek V4 ProDeepSeek55.4%max effortDeepSeek report3 values · 1 reportApr 26, 2026 · source
6Gemini 3.1 ProGoogle54.2%high effortDeepSeek report1 value · 1 reportApr 26, 2026 · source
7DeepSeek-V4-FlashDeepSeek52.6%max effortDeepSeek report3 values · 1 reportApr 26, 2026 · source

Effort curve

Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.

0
No cost-linked effort sweep for this version.
Definition and comparison boundarypublic methodology

Can the model complete substantial programming work under this benchmark's agent setup?

Software implementation, debugging, or repository work under the published evaluation protocol. Higher is better. The value is the percentage reported in this lab's table.

Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.

Frontier comparison and across-modes table.

Method / source