SWE-bench Multilingual

Multilingual code issue resolution.

298Public tasks

96.4%Highest score

The SWE-bench Multilingual leaderboard

A multilingual expansion for SWE-bench that tests whether LLMs can resolve 298 real-world GitHub issues in 42 repositories spanning nine programming languages.

Paper Data Code Sample task

Model

Score

DeepSeek-V4-Pro-0813Max

96.4%

Opus 5Max

96.2%

DeepSeek-V4-FlashMax

95.4%

Fable 5Max

94.8%

GPT-5.6 TerraMax

94.4%

50%

60%

70%

80%

90%

100%

Pass@1

SWE-bench Multilingual

The SWE-bench Multilingual leaderboard

The latest on frontier AI performance, straight to your inbox.