Skip to content

SWE Multilingual

A multilingual software-engineering benchmark for real-world code issue resolution across multiple programming languages.

46 models

Higher is better.

  1. 1Claude Opus 5.5Anthropic93.9
  2. 2Claude Mythos 5Anthropic92.2
  3. 3Claude Sonnet 5.5Anthropic90.3
  4. 4Claude Opus 5Anthropic89.5
  5. 5Claude Fable 5.1Anthropic89.1
  6. 6Ling 3.1 FlashInclusionAI88.5
  7. 7Claude Opus 4.8Anthropic84.4
  8. 8Hy4 previewTencent82.9
  9. 9Qwen3.8-Flash-NextAlibaba81.0
  10. 10Qwen3.8-Omni-FlashAlibaba80.5
  11. 11Composer 2.5Cursor79.8
  12. 12Ornith-1.5-397BOrnith AI79.6
  13. 13Ornith-1.0-397BDeepReinforce AI78.9
  14. 14Laguna S 2.1Poolside78.5
  15. 15Claude Sonnet 5Anthropic78.3
  16. 16Qwen3.7 MaxAlibaba78.3
  17. 17Beam 1.0 501B-A23BReflection AI78.3
  18. 18Grok 4.5xAI78.0
  19. 19SWE-1.7Cognition77.8
  20. 20Claude Opus 4.5Anthropic77.5
  21. 21Kimi K2.6Moonshot AI76.7
  22. 22MiniMax M2.7MiniMax76.5
  23. 23DeepSeek V4 Pro 0813DeepSeek76.2
  24. 24Qwen3.7 PlusAlibaba75.8
  25. 25dots3-note PreviewDots Studio75.7
  26. 26Qwen3.6 PlusAlibaba73.8
  27. 27Composer 2Cursor73.7
  28. 28GLM-5Z.AI73.3
  29. 29DeepSeek V4 Flash 0731DeepSeek73.3
  30. 30Kimi K2.5Moonshot AI73.0
  31. 31Ling 3.0 FlashInclusionAI72.4
  32. 32MiMo-V2-FlashXiaomi71.7
  33. 33Ornith-1.5-35B-A3BOrnith AI71.4
  34. 34Qwen3.6-27BAlibaba71.3
  35. 35Qwen3.5 397BAlibaba69.3
  36. 36Nemotron 3.5 Super VL 120B A12BNVIDIA67.7
  37. 37Nemotron 3 UltraNVIDIA67.7
  38. 38Qwen3.6-35B-A3BAlibaba67.2
  39. 39Laguna M.1Poolside63.1
  40. 40LongCat-Flash-Lite-SparseMeituan59.3
Higher is better. Showing the top 40 of 46. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Repository task completion
Tasks
Multilingual software-engineering tasks
Authors
MiniMax (2026)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.

All results

1Claude Opus 5.5 Anthropic93.9
2Claude Mythos 5 Anthropic92.2
3Claude Sonnet 5.5 Anthropic90.3
4Claude Opus 5 Anthropic89.5
5Claude Fable 5.1 Anthropic89.1
6Ling 3.1 Flash InclusionAI88.5
7Claude Opus 4.8 Anthropic84.4
8Hy4 preview Tencent82.9
9Qwen3.8-Flash-Next Alibaba81.0
10Qwen3.8-Omni-Flash Alibaba80.5
11Composer 2.5 Cursor79.8
12Ornith-1.5-397B Ornith AI79.6
13Ornith-1.0-397B DeepReinforce AI78.9
14Laguna S 2.1 Poolside78.5
15Claude Sonnet 5 Anthropic78.3
16Qwen3.7 Max Alibaba78.3
17Beam 1.0 501B-A23B Reflection AI78.3
18Grok 4.5 xAI78.0
19SWE-1.7 Cognition77.8
20Claude Opus 4.5 Anthropic77.5
21Kimi K2.6 Moonshot AI76.7
22MiniMax M2.7 MiniMax76.5
23DeepSeek V4 Pro 0813 DeepSeek76.2
24Qwen3.7 Plus Alibaba75.8
25dots3-note Preview Dots Studio75.7
26Qwen3.6 Plus Alibaba73.8
27Composer 2 Cursor73.7
28GLM-5 Z.AI73.3
29DeepSeek V4 Flash 0731 DeepSeek73.3
30Kimi K2.5 Moonshot AI73.0
31Ling 3.0 Flash InclusionAI72.4
32MiMo-V2-Flash Xiaomi71.7
33Ornith-1.5-35B-A3B Ornith AI71.4
34Qwen3.6-27B Alibaba71.3
35Qwen3.5 397B Alibaba69.3
36Nemotron 3.5 Super VL 120B A12B NVIDIA67.7
37Nemotron 3 Ultra NVIDIA67.7
38Qwen3.6-35B-A3B Alibaba67.2
39Laguna M.1 Poolside63.1
40LongCat-Flash-Lite-Sparse Meituan59.3
41Qwen3-Coder 480B-A35B Alibaba54.7
42Ornith-1.5-9B Ornith AI54.4
43Kimi K2 Moonshot AI47.3
44Nemotron 3.5 Lightning 30B A3B NVFP4 NVIDIA36.5
45Granite 4.2 8B IBM30.8
46LLaDA2.2-flash InclusionAI25.0