benchgap
Coding

SWE-Rebench leaderboard

As of 2026-11-21, the highest measured score on SWE-Rebench is 65.3% by Claude Opus 4.6. 125 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Opus 4.665.3%measured
2GLM-562.8%measured
3GLM-5.162.7%measured
4Gemini 4 Argon62.3%estimated ± 6.5 pp, low confidence
5Mistral Large 462.2%estimated ± 6.5 pp, low confidence
6GPT-5.562.2%estimated ± 6.5 pp, low confidence
7GPT-5.462.2%estimated ± 6.5 pp, low confidence
8Claude Opus 4.6 (Adaptive)62.1%estimated ± 6.5 pp, low confidence
9Gemini 3.5 Flash62.0%estimated ± 6.5 pp, low confidence
10GPT-5.4 mini62.0%estimated ± 6.5 pp, low confidence
11Claude Fable 561.8%estimated ± 3.6 pp, low confidence
12Claude Mythos 561.8%estimated ± 3.6 pp, low confidence
13Claude Opus 561.8%estimated ± 3.6 pp, low confidence
14Ember-161.8%estimated ± 3.6 pp, low confidence
15Claude Opus 4.861.8%estimated ± 3.6 pp, low confidence
16Claude Opus 4.761.8%estimated ± 3.6 pp, low confidence
17Ornith-1.5-397B61.8%estimated ± 3.6 pp, low confidence
18Claude Sonnet 561.8%estimated ± 3.6 pp, low confidence
19Ornith-1.0-397B61.8%estimated ± 3.6 pp, medium confidence
20Inkling 2 Small61.8%estimated ± 3.6 pp, medium confidence
21Beam 1.0 501B-A23B61.8%estimated ± 3.6 pp, medium confidence
22Claude Opus 4.561.8%estimated ± 3.6 pp, medium confidence
23Beam 1.0 501B-A23B NVFP461.8%estimated ± 3.6 pp, medium confidence
24DeepSeek V4 Pro61.8%estimated ± 3.6 pp, medium confidence
25MiniMax M361.8%estimated ± 3.6 pp, medium confidence
26Qwen3.7 Max61.8%estimated ± 3.6 pp, medium confidence
27Inkling-Small61.8%estimated ± 3.6 pp, medium confidence
28Kimi K2.661.8%estimated ± 3.6 pp, medium confidence
29Beam 1.0 501B-A23B FP861.8%estimated ± 3.6 pp, medium confidence
30GPT-5.261.8%estimated ± 3.6 pp, medium confidence
31Inkling 2 Mini61.7%estimated ± 3.6 pp, medium confidence
32DeepSeek V4 Flash 073161.7%estimated ± 3.6 pp, medium confidence
33Ornith-1.5-35B-A3B61.7%estimated ± 3.6 pp, medium confidence
34Qwen3.6 Plus61.7%estimated ± 3.6 pp, medium confidence
35BTL-461.6%estimated ± 3.6 pp, medium confidence
36dots3-note Preview61.6%estimated ± 3.6 pp, medium confidence
37MiMo-V2-Pro61.3%estimated ± 3.6 pp, medium confidence
38Apodex 1.161.0%estimated ± 3.6 pp, medium confidence
39Qwen3.7 Plus61.0%estimated ± 3.6 pp, medium confidence
40GPT-5.2-Codex61.0%estimated ± 6.5 pp, low confidence
41DeepSeek V3.260.9%measured
42Inkling60.9%estimated ± 3.6 pp, medium confidence
43Mistral Medium 3.5 128B60.9%estimated ± 3.6 pp, medium confidence
44Claude Sonnet 4.660.7%measured
45Muse Spark60.6%estimated ± 3.6 pp, medium confidence
46Claude Sonnet 4.560.2%estimated ± 3.6 pp, medium confidence
47Qwen3.6-27B60.2%estimated ± 3.6 pp, medium confidence
48Grok 4.2059.0%estimated ± 3.6 pp, medium confidence
49Gemini 3.1 Pro58.9%estimated ± 6.5 pp, low confidence
50Qwen3.5-27B58.9%measured
51GLM-4.758.7%measured
52Ripple-65B Preview58.6%estimated ± 3.6 pp, medium confidence
53Kimi K2.558.5%measured
54GPT-5.3 Codex58.2%measured
55Qwen3.8-27B58.1%estimated ± 8.9 pp, low confidence
56Quasar 438B58.1%estimated ± 8.9 pp, low confidence
57Apodex 1.1 Mini58.1%estimated ± 8.9 pp, low confidence
58Kimi K2.7 Code58.1%estimated ± 8.9 pp, low confidence
59MiMo-V2.5-Pro58.1%estimated ± 8.9 pp, low confidence
60Hy358.1%estimated ± 8.9 pp, low confidence
61Ling 3.0 Flash58.1%estimated ± 8.9 pp, low confidence
62Gemini 3.5 Flash-Lite58.1%estimated ± 8.9 pp, low confidence
63Qwen3.5 397B58.0%estimated ± 3.6 pp, medium confidence
64Composer 258.0%measured
65Muse Glimmer 30B57.8%estimated ± 3.6 pp, medium confidence
66Ornith-1.0-35B57.4%estimated ± 3.6 pp, medium confidence
67MiMo-V2-Omni57.1%estimated ± 3.6 pp, medium confidence
68Laguna M.157.1%estimated ± 3.6 pp, medium confidence
69Claude 4.1 Opus57.1%estimated ± 3.6 pp, medium confidence
70Hy3 Preview57.1%estimated ± 3.6 pp, medium confidence
71Step 3.5 Flash57.1%estimated ± 3.6 pp, medium confidence
72Flint57.1%estimated ± 3.6 pp, medium confidence
73MAI-Thinking-157.0%estimated ± 3.6 pp, medium confidence
74MiMo-V2-Flash57.0%estimated ± 3.6 pp, medium confidence
75Qwen3.6-35B-A3B57.0%estimated ± 3.6 pp, medium confidence
76Claude Haiku 4.557.0%estimated ± 3.6 pp, medium confidence
77Jais-3-MoVA-36B-A4B57.0%estimated ± 3.6 pp, medium confidence
78Claude 4 Sonnet57.0%estimated ± 3.6 pp, medium confidence
79MAI-Code-1.1-Flash57.0%estimated ± 3.6 pp, medium confidence
80Claude 4 Opus57.0%estimated ± 3.6 pp, medium confidence
81Jais-3-32B57.0%estimated ± 3.6 pp, medium confidence
82Qwen3.5-122B-A10B57.0%estimated ± 3.6 pp, medium confidence
83Nemotron 3 Ultra57.0%estimated ± 3.6 pp, medium confidence
84Nemotron 3.5 Super VL 120B A12B57.0%estimated ± 3.6 pp, medium confidence
85Nemotron 3.5 Super VL 120B A12B NVFP457.0%estimated ± 3.6 pp, medium confidence
86Laguna XS 2.157.0%estimated ± 3.6 pp, medium confidence
87Grok Code Fast 157.0%estimated ± 3.6 pp, medium confidence
88Ornith-1.5-9B57.0%estimated ± 3.6 pp, medium confidence
89Solar Pro 457.0%estimated ± 3.6 pp, medium confidence
90Solar Open 257.0%estimated ± 3.6 pp, medium confidence
91Jais-3-375B-A23B57.0%estimated ± 3.6 pp, medium confidence
92Jais-3-7B57.0%estimated ± 3.6 pp, medium confidence
93Laguna XS.257.0%estimated ± 3.6 pp, medium confidence
94Qwen3-Coder 480B-A35B57.0%estimated ± 3.6 pp, medium confidence
95Ornith-1.0-9B57.0%estimated ± 3.6 pp, medium confidence
96K-EXAONE 2.057.0%estimated ± 3.6 pp, low confidence
97LongCat-Flash-Lite-Sparse57.0%estimated ± 3.6 pp, low confidence
98DeepSeek V3.157.0%estimated ± 3.6 pp, low confidence
99Kimi K257.0%estimated ± 3.6 pp, low confidence
100Gemini 2.5 Pro57.0%estimated ± 3.6 pp, low confidence
101Spark-X2.6-4B57.0%estimated ± 3.6 pp, low confidence
102GPT-OSS 120B57.0%estimated ± 3.6 pp, low confidence
103Claude 3.7 Sonnet57.0%estimated ± 3.6 pp, low confidence
104Ternary Bonsai 2 27B57.0%estimated ± 3.6 pp, low confidence
105GPT-OSS 20B57.0%estimated ± 3.6 pp, low confidence
106Nemotron 3 Super 100B57.0%estimated ± 3.6 pp, low confidence
107DeepSeek-R1-052857.0%estimated ± 3.6 pp, low confidence
108GLM-4.7-Flash57.0%estimated ± 3.6 pp, low confidence
109GPT-4.157.0%estimated ± 3.6 pp, low confidence
110Granite 4.2 30B57.0%estimated ± 3.6 pp, low confidence
111MiniMax M1 80k57.0%estimated ± 3.6 pp, low confidence
112Qwen3.5-35B-A3B53.7%measured
113GPT-5.4 nano52.2%estimated ± 6.5 pp, low confidence
114MiniMax M2.751.9%measured
115GPT-5.148.8%estimated ± 6.5 pp, low confidence
116Claude Sonnet 4.5 Thinking42.8%estimated ± 6.5 pp, low confidence
117Gemma 4 31B41.6%measured
118GPT-5.1-Codex-Max41.1%estimated ± 6.5 pp, low confidence
119Claude Opus 4.5 Thinking34.7%estimated ± 6.5 pp, low confidence
120Gemini 3 Flash32.8%estimated ± 6.5 pp, low confidence
121GPT-532.3%estimated ± 6.5 pp, low confidence
122Qwen3.5 Plus12.4%estimated ± 6.5 pp, low confidence
123Grok 4.311.2%estimated ± 8.9 pp, low confidence
124MiniMax M2.59.3%estimated ± 6.5 pp, low confidence
125Gemini 3 Pro7.6%estimated ± 6.5 pp, low confidence
126GPT-5 mini7.3%estimated ± 6.5 pp, low confidence
127GPT-5.1-Codex4.8%estimated ± 6.5 pp, low confidence
128Claude Haiku 4.5 Thinking2.2%estimated ± 6.5 pp, low confidence
129o10.1%estimated ± 8.9 pp, low confidence
130Step 3.7 Flash0.1%estimated ± 8.9 pp, low confidence
131Gemma 4 26B A4B0.0%estimated ± 8.9 pp, low confidence
132Qwen3 Max0.0%estimated ± 6.5 pp, low confidence
133GLM-4.60.0%estimated ± 6.5 pp, low confidence
134Grok 4.1 Fast (Reasoning)0.0%estimated ± 6.5 pp, low confidence
135o1-preview0.0%estimated ± 8.9 pp, low confidence
136K-Exaone0.0%estimated ± 8.9 pp, low confidence
137Gemini 3.1 Flash-Lite0.0%estimated ± 6.5 pp, low confidence
138Grok 4 Fast (Reasoning)0.0%estimated ± 6.5 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General