DeepSWE Leaderboard

Benchmark Model Rank Results
code-generation-on-deepsweClaude Fable 5 (xhigh) (MiniSWE-Agent)#10Pass@1: 0.6991
code-generation-on-deepsweQwen3.8 Max (xhigh) (MiniSWE-Agent)#20Pass@1: 0.5746
code-generation-on-deepsweClaude Opus 4.8 (max) (MiniSWE-Agent)–Pass@1: 0.5897
code-generation-on-deepsweClaude Sonnet 4.6 (high) (MiniSWE-Agent)–Pass@1: 0.2993
code-generation-on-deepsweClaude Sonnet 5 (max) (MiniSWE-Agent)–Pass@1: 0.5385
code-generation-on-deepsweGPT-5.4 (xhigh) (MiniSWE-Agent)–Pass@1: 0.5177
code-generation-on-deepsweGPT-5.5 (xhigh) (MiniSWE-Agent)–Pass@1: 0.6704
code-generation-on-deepsweGemini 3.1 Pro Preview (high) (MiniSWE-Agent)–Pass@1: 0.1173
code-generation-on-deepsweGemini 3.5 Flash (high) (MiniSWE-Agent)–Pass@1: 0.3606
code-generation-on-deepsweGemini 3.6 Flash (high) (MiniSWE-Agent)–Pass@1: 0.4668
code-generation-on-deepsweGrok 4.6 (medium) (MiniSWE-Agent)–Pass@1: 0.6748
code-generation-on-deepsweKimi K2.7 Code (MiniSWE-Agent)–Pass@1: 0.3053
code-generation-on-deepsweMuse Spark 1.2 (xhigh) (MiniSWE-Agent)–Pass@1: 0.5487