benchgap
Coding

FrontierSWE v2 leaderboard

As of 2026-11-21, the highest measured score on FrontierSWE v2 is 69.9% by Claude Fable 5.2. 154 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Fable 5.269.9%measured
2GPT-6 Astra65.5%measured
3Gemini 4 Neon62.4%measured
4Claude Opus 5.562.3%measured
5Claude Sonnet 5.561.9%measured
6Gemini 4 Carbon61.2%measured
7GPT-6.2 Astra60.6%estimated ± 8.0 pp, medium confidence
8GPT-6.2 Sol60.2%estimated ± 8.0 pp, medium confidence
9Claude Fable 5.156.3%measured
10GPT-6.1 Sol56.0%estimated ± 8.0 pp, medium confidence
11Gemini 4 Argon55.0%measured
12Gemini 4 Hydrogen53.1%measured
13Claude Opus 552.0%measured
14GPT-6 Sol51.1%estimated ± 8.0 pp, medium confidence
15MiniMax M3.1 Flash50.8%estimated ± 10.4 pp, low confidence
16Muse Wonder Full50.7%estimated ± 11.1 pp, low confidence
17Claude Mythos 550.7%estimated ± 8.3 pp, medium confidence
18Muse Wonder47.6%estimated ± 5.8 pp, medium confidence
19Claude Fable 547.0%measured
20Hy446.7%estimated ± 8.0 pp, medium confidence
21Claude Haiku 5.543.8%measured
22Qwen4 Max42.3%estimated ± 11.2 pp, low confidence
23DeepSeek V4.1 Flash 101742.0%estimated ± 5.8 pp, medium confidence
24Sakana Fugu-Ultra39.3%estimated ± 8.3 pp, medium confidence
25Step 5 Preview39.2%estimated ± 14.2 pp, low confidence
26SWE-238.3%estimated ± 8.0 pp, medium confidence
27SWE-2 Flash38.2%estimated ± 8.0 pp, medium confidence
28GPT-5.537.6%estimated ± 8.0 pp, medium confidence
29GPT-6 Luna36.5%estimated ± 8.0 pp, medium confidence
30Kimi K3.1 Pro32.7%estimated ± 8.0 pp, medium confidence
31Mistral Large 432.4%estimated ± 8.0 pp, medium confidence
32GPT-5.6 Sol32.2%measured
33MiMo-V2.6-Pro31.0%estimated ± 11.1 pp, low confidence
34Ling 3.1 Flash30.7%estimated ± 8.3 pp, medium confidence
35GLM-5.330.2%measured
36Grok 4.729.5%measured
37GPT-6.1 Terra29.5%estimated ± 14.2 pp, low confidence
38GPT-5.6 Terra28.6%estimated ± 5.8 pp, medium confidence
39SWE-1.726.4%estimated ± 8.0 pp, low confidence
40Claude Opus 4.826.1%estimated ± 8.0 pp, low confidence
41Beam 1.0 501B-A23B26.1%estimated ± 8.3 pp, medium confidence
42Kimi K325.9%measured
43Inkling 2 Mini25.8%estimated ± 8.0 pp, low confidence
44Muse Spark 1.325.8%measured
45Beam 1.0 501B-A23B FP825.8%estimated ± 8.3 pp, medium confidence
46Hy4 preview25.4%estimated ± 8.3 pp, medium confidence
47Grok 4.625.3%measured
48Beam 1.0 501B-A23B NVFP425.3%estimated ± 8.3 pp, medium confidence
49Inkling 2 Small25.1%measured
50Ornith-1.5-397B24.4%estimated ± 8.3 pp, medium confidence
51Grok 4.523.7%estimated ± 8.3 pp, medium confidence
52Claude Opus 4.723.0%estimated ± 8.3 pp, medium confidence
53Qwen3.8 Max Preview22.6%estimated ± 14.2 pp, low confidence
54Qwen3.8-Omni-Flash21.3%estimated ± 8.3 pp, medium confidence
55MiMo-V2.6-Flash21.0%estimated ± 14.2 pp, low confidence
56Gemini 3.7 Flash20.3%measured
57Qwen3.8-Flash-Next19.9%estimated ± 8.3 pp, medium confidence
58Gemini 3.8 Flash19.6%measured
59Ornith-1.0-397B19.4%estimated ± 8.3 pp, medium confidence
60GLM-5.219.2%estimated ± 8.3 pp, medium confidence
61Qwen3.8-27B18.5%estimated ± 8.3 pp, medium confidence
62Muse Spark 1.118.2%estimated ± 8.3 pp, medium confidence
63GLM-5.3-Flash18.1%measured
64Ripple-65B Preview18.0%estimated ± 8.3 pp, medium confidence
65Sage-A1.517.7%estimated ± 8.3 pp, medium confidence
66GPT-5.6 Luna17.3%estimated ± 5.8 pp, low confidence
67dots3-note Preview17.3%estimated ± 8.3 pp, medium confidence
68Qwen3.7 Max16.6%estimated ± 8.3 pp, medium confidence
69Hy315.9%estimated ± 14.2 pp, low confidence
70Hy3 Preview15.9%estimated ± 14.2 pp, low confidence
71Qwen3.8 Max15.8%measured
72Quasar 438B15.1%estimated ± 14.2 pp, low confidence
73Solar Pro 414.9%estimated ± 14.2 pp, low confidence
74Atria Dawn Preview14.9%estimated ± 8.3 pp, medium confidence
75Ornith-1.5-35B-A3B14.9%estimated ± 8.3 pp, medium confidence
76Laguna S 2.114.5%estimated ± 8.3 pp, medium confidence
77MiniMax M313.9%estimated ± 8.3 pp, medium confidence
78Sakana Fugu13.9%estimated ± 8.3 pp, medium confidence
79Claude Sonnet 513.6%estimated ± 5.8 pp, low confidence
80Kimi K2.613.2%estimated ± 8.3 pp, medium confidence
81GLM-5.112.8%estimated ± 8.3 pp, medium confidence
82Gemini 2.5 Pro12.3%estimated ± 14.2 pp, low confidence
83Muse Spark Open12.0%estimated ± 8.0 pp, low confidence
84Muse Spark 1.212.0%measured
85GPT-5.411.6%estimated ± 8.3 pp, medium confidence
86Qwen3.7 Plus11.4%estimated ± 8.3 pp, medium confidence
87Apodex 1.111.2%estimated ± 14.2 pp, low confidence
88Apodex 1.1 Mini11.2%estimated ± 14.2 pp, low confidence
89Gemma 4 31B11.2%estimated ± 14.2 pp, low confidence
90Qwen 3.6 Max (preview)10.9%estimated ± 8.3 pp, medium confidence
91MiMo-V2.5-Pro10.7%estimated ± 8.3 pp, medium confidence
92Claude Opus 4.510.6%estimated ± 8.3 pp, medium confidence
93Gemini 3.6 Flash10.1%estimated ± 9.5 pp, medium confidence
94GPT-5.3 Codex10.1%estimated ± 8.3 pp, medium confidence
95Ling 3.0 Flash9.7%estimated ± 8.3 pp, medium confidence
96Qwen3.6 Plus9.7%estimated ± 8.3 pp, medium confidence
97Gemini 3.1 Pro9.7%estimated ± 9.5 pp, medium confidence
98Ling 3.0 Flash VL9.6%estimated ± 14.2 pp, low confidence
99Claude Opus 4.6 (Adaptive)9.4%estimated ± 9.5 pp, medium confidence
100Kimi K2.7 Code9.4%estimated ± 9.5 pp, medium confidence
101Step 3.7 Flash9.2%estimated ± 8.3 pp, medium confidence
102Claude Sonnet 4.69.0%estimated ± 9.5 pp, low confidence
103MiniMax M2.79.0%estimated ± 8.3 pp, medium confidence
104MiMo-V2.58.8%estimated ± 8.3 pp, medium confidence
105Claude Opus 4.5 Thinking8.6%estimated ± 9.5 pp, low confidence
106Gemini 3 Pro8.6%estimated ± 9.5 pp, low confidence
107Inkling-Small8.5%estimated ± 8.3 pp, medium confidence
108Gemini 3 Flash8.0%estimated ± 9.5 pp, low confidence
109GPT-5.28.0%estimated ± 8.3 pp, medium confidence
110DeepSeek V4 Pro7.6%estimated ± 8.3 pp, medium confidence
111GPT-5.4 mini7.3%estimated ± 9.5 pp, low confidence
112K-EXAONE 2.07.2%estimated ± 14.2 pp, low confidence
113Gemini 3.5 Flash7.1%estimated ± 8.3 pp, medium confidence
114GLM-57.1%estimated ± 8.3 pp, medium confidence
115GPT-5.2-Codex7.1%estimated ± 9.5 pp, low confidence
116Grok 4.36.8%estimated ± 9.5 pp, low confidence
117Claude Sonnet 4.5 Thinking6.4%estimated ± 9.5 pp, low confidence
118GPT-5.16.3%estimated ± 9.5 pp, low confidence
119GPT-5.4 nano6.3%estimated ± 9.5 pp, low confidence
120A.X K26.3%estimated ± 14.2 pp, low confidence
121GLM-4.76.2%estimated ± 9.5 pp, low confidence
122GPT-56.1%estimated ± 9.5 pp, low confidence
123Nemotron 3 Ultra6.1%estimated ± 9.5 pp, low confidence
124Trinity-Large-Preview5.9%estimated ± 14.2 pp, low confidence
125Trinity-Large-Thinking5.9%estimated ± 14.2 pp, low confidence
126Spark-X2.6-4B5.7%estimated ± 8.3 pp, medium confidence
127Gemini 3.5 Flash-Lite5.6%estimated ± 8.3 pp, low confidence
128Claude Haiku 4.55.5%estimated ± 9.5 pp, low confidence
129Mistral Medium 3.5 128B5.5%estimated ± 9.5 pp, low confidence
130Gemma 4 26B A4B5.5%estimated ± 14.2 pp, low confidence
131Nemotron 3.5 Bolt 74B-A9B5.2%estimated ± 14.2 pp, low confidence
132Qwen3.5-122B-A10B5.2%estimated ± 14.2 pp, low confidence
133Qwen3.5 Flash5.0%estimated ± 9.5 pp, low confidence
134Gemini 3.1 Flash-Lite4.7%estimated ± 9.5 pp, low confidence
135DeepSeek V3 03244.7%estimated ± 14.2 pp, low confidence
136GPT-OSS 20B4.6%estimated ± 14.2 pp, low confidence
137Mistral Small 44.6%estimated ± 14.2 pp, low confidence
138North Mini Code4.6%estimated ± 14.2 pp, low confidence
139Command A+4.4%estimated ± 14.2 pp, low confidence
140Mercury 2.54.4%estimated ± 14.2 pp, low confidence
141Qwen3.6-27B4.4%estimated ± 8.3 pp, low confidence
142Claude Opus 4.64.2%estimated ± 8.3 pp, low confidence
143Inkling4.1%measured
144MiMo-V2-Flash3.6%estimated ± 14.3 pp, low confidence
145Mistral Large 33.3%estimated ± 14.2 pp, low confidence
146MAI-Thinking-13.1%estimated ± 8.3 pp, low confidence
147DeepSeek V32.9%estimated ± 14.2 pp, low confidence
148DeepSeek V4 Flash 07312.8%estimated ± 8.3 pp, low confidence
149Muse Spark2.5%estimated ± 8.3 pp, low confidence
150GPT-OSS 120B2.1%estimated ± 14.2 pp, low confidence
151Nemotron 3.5 Lightning 30B A3B NVFP41.5%estimated ± 14.2 pp, low confidence
152Grok 4.201.4%estimated ± 8.3 pp, low confidence
153Llama 4 Maverick1.4%estimated ± 14.2 pp, low confidence
154Nemotron 3.5 Super VL 120B A12B1.4%estimated ± 8.3 pp, low confidence
155o10.9%estimated ± 14.3 pp, low confidence
156Mistral Small 3.10.6%estimated ± 14.2 pp, low confidence
157Nemotron 3.5 Super VL 120B A12B NVFP40.5%estimated ± 8.3 pp, low confidence
158MiniCPM5-2B0.5%estimated ± 14.2 pp, low confidence
159o1-preview0.4%estimated ± 14.3 pp, low confidence
160Muse Glimmer 30B0.4%estimated ± 8.3 pp, low confidence
161K-Exaone0.3%estimated ± 14.3 pp, low confidence
162Composer 2.50.2%estimated ± 5.8 pp, low confidence
163Gemma 4 12B0.2%estimated ± 14.3 pp, low confidence
164Ling 2.6 Flash0.1%estimated ± 14.3 pp, low confidence
165Gemini 1.5 Pro0.0%estimated ± 14.3 pp, low confidence
166Granite 4.2 30B0.0%estimated ± 8.3 pp, low confidence
167Granite 4.2 8B0.0%estimated ± 8.3 pp, low confidence
168Kimi K2.50.0%estimated ± 8.3 pp, low confidence
169Laguna M.10.0%estimated ± 8.3 pp, low confidence
170Laguna XS.20.0%estimated ± 8.3 pp, low confidence
171Laguna XS 2.10.0%estimated ± 8.3 pp, low confidence
172LLaDA2.2-flash0.0%estimated ± 8.3 pp, low confidence
173LongCat-Flash-Lite-Sparse0.0%estimated ± 8.3 pp, low confidence
174Mellum2.1-12B-A2.5B-Thinking0.0%estimated ± 8.3 pp, low confidence
175Ornith-1.0-35B0.0%estimated ± 8.3 pp, low confidence
176Ornith-1.0-9B0.0%estimated ± 8.3 pp, low confidence
177Ornith-1.5-9B0.0%estimated ± 8.3 pp, low confidence
178Qwen3.5 397B0.0%estimated ± 8.3 pp, low confidence
179Qwen3.6-35B-A3B0.0%estimated ± 8.3 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General