← SWE-bench leaderboards

Lite

SWE-bench · 50 models

RankModelScore
1ExpeRepair-v1.0 + Claude 4 Sonnet60.33
2Refact.ai Agent60
3KGCompass + Claude 4 Sonnet (20250514)58.33
4SWE-agent + Claude 4 Sonnet56.67
5Isoform55
6SemAgent_Multi-v1.051.67
7Isea51.33
8EntroPO + R2E + Qwen3-Coder-30B-A3B-Instruct49.67
9Blackbox AI Agent49
9Codev49
11Gru(2024-12-08)48.67
12ExpeRepair-v1.048.33
12Globant Code Fixer Agent48.33
14SWE-agent + Claude 3.7 Sonnet48
15devlo47.33
16DARS Agent47
17KGCompass + Claude 3.5 Sonnet (20241022)46
18EntroPO + R2E + Qwen3-Coder-30B-A3B-Instruct45
19Kodu-v1 + Claude-3.5 Sonnet (20241022)44.67
20CodeFuse-CGM44
21CodeStory Aide + Mixed Models43
22Lingxi42.67
23Codart AI41.67
23OpenHands + CodeAct v2.1 (claude-3-5-sonnet-20241022)41.67
25PatchKitty-0.9 + Claude-3.5 Sonnet (20241022)41.33
26Composio SWE-Kit (2024-10-30)41
26OrcaLoca + Agentless-1.5 + Claude-3.5 Sonnet (20241022)41
28Agentless-1.5 + Claude-3.5 Sonnet (20241022)40.67
29OpenCSG Starship Agentic Coder + GPT 4 (0806)39.67
30Bytedance MarsCode Agent39.33
31Moatless Tools + Claude 3.5 Sonnet (20241022)39
32Honeycomb38.33
32Moatless Tools + Claude 3.5 Sonnet (20241022)38.33
34AbanteAI MentatBot + GPT 4o (2024-05-13)38
35Patched.Codes Patchwork37
36KGCompass + DeepSeek V336.67
37AppMap Navie v236
38CodeFuse-AAIS35.67
38Gru(2024-08-11)35.67
40Isoform35
41Bytedance MarsCode Agent + GPT 4o (2024-05-13)34
41SuperCoder2.034
43Alibaba Lingma Agent33
44Agentless Lite + O3 Mini (20250214)32.33
45Agentless-1.5 + GPT 4o (2024-05-13)32
46CodeShellTester + GPT 4o (2024-05-13)31.33
46Factory Code Droid31.33
48AutoCodeRover (v20240620) + GPT 4o (2024-05-13)30.67
48Moatless Tools + Deepseek V330.67
50Aegis - o3-mini_1.030.33