Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Agentic Software Engineering on SWE-bench Multilingual (SWE-M)
Loading...
50
Success Rate
Claude-Opus-4.5
-2
11.5
25
38.5
Jun 16, 2026
Success Rate
Updated 1mo ago
Evaluation Results
Method
Method
Links
Success Rate
Claude-Opus-4.5
2026.06
50
Gemini-3-Pro
2026.06
42.7
DeepSeek-V3.2
2026.06
37.4
Kimi-K2-Instruct-0905
2026.06
33.5
Qwen3-Coder-480B-A35B-Instruct
2026.06
32.7
LoopCoder-v2
R (number of refinemen...
2026.06
31
LoopCoder-v2
R (number of refinemen...
2026.06
14
LoopCoder-v2
R (number of refinemen...
2026.06
11
LoopCoder-v2
R (number of refinemen...
2026.06
9.3
DeepSeek-Coder-V2-Lite-Instruct
2026.06
0
Qwen2.5-Coder-7B-Instruct
2026.06
0
Seed-Coder-8B-Instruct
2026.06
0
Qwen2.5-Coder-14B-Instruct
2026.06
0
Qwen2.5-Coder-32B-Instruct
2026.06
0
Feedback
Search any
task
Search any
task