benchgap
Coding

SciCode leaderboard

As of 2026-11-21, the highest measured score on SciCode is 60.1% by Sakana Fugu. 226 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Fable 5.267.0%estimated ± 3.1 pp, medium confidence
2Claude Opus 5.566.2%estimated ± 3.1 pp, medium confidence
3Qwen4 Max62.2%estimated ± 7.6 pp, low confidence
4Muse Wonder Full62.1%estimated ± 3.1 pp, medium confidence
5Hy461.8%estimated ± 7.6 pp, low confidence
6Pareto 26.961.4%estimated ± 7.6 pp, low confidence
7Gemini 4 Argon61.3%estimated ± 3.1 pp, medium confidence
8MAI Code 261.0%estimated ± 7.6 pp, low confidence
9SWE-261.0%estimated ± 7.6 pp, low confidence
10Particle-160.8%estimated ± 7.6 pp, low confidence
11Claude Sonnet 5.560.5%estimated ± 3.1 pp, medium confidence
12MiMo-V2.6-Pro60.4%estimated ± 3.1 pp, medium confidence
13Kimi K3.1 Pro60.3%estimated ± 3.1 pp, medium confidence
14ERNIE 5.5 Flash Preview60.1%estimated ± 7.6 pp, low confidence
15Sakana Fugu60.1%measured
16Muse Wonder59.8%estimated ± 3.1 pp, medium confidence
17Pareto 26.10 Preview59.5%estimated ± 7.6 pp, low confidence
18Step 5 Preview58.9%measured
19Sakana Fugu-Ultra58.7%measured
20Muse Spark 1.358.4%estimated ± 3.1 pp, high confidence
21Muse Spark Open58.0%estimated ± 3.1 pp, high confidence
22Claude Fable 5.157.8%estimated ± 2.9 pp, low confidence
23GPT-6 Sol57.2%estimated ± 3.1 pp, high confidence
24Grok 4.757.0%estimated ± 3.1 pp, high confidence
25Muse Spark 1.257.0%estimated ± 3.1 pp, high confidence
26Hy4 flash57.0%estimated ± 7.6 pp, low confidence
27LongCat-2.557.0%estimated ± 7.6 pp, low confidence
28Claude Fable 556.4%estimated ± 2.9 pp, low confidence
29Qwen3.8-Omni-Flash56.3%estimated ± 4.9 pp, high confidence
30GPT-6 Astra56.2%estimated ± 3.1 pp, high confidence
31Gemini 3.8 Flash55.8%estimated ± 2.9 pp, low confidence
32Ultramarine 1 Flash55.5%estimated ± 7.6 pp, low confidence
33SWE-2 Flash54.8%estimated ± 7.6 pp, low confidence
34GPT-6.1 Terra54.8%estimated ± 3.1 pp, high confidence
35Claude Opus 554.8%estimated ± 2.9 pp, low confidence
36MAI Code 2 Flash54.7%estimated ± 7.6 pp, low confidence
37Claude Haiku 5.554.7%estimated ± 3.1 pp, high confidence
38Claude Mythos 554.3%estimated ± 3.6 pp, medium confidence
39GPT-6 Luna54.3%estimated ± 3.1 pp, high confidence
40Gemini 3.7 Flash54.2%estimated ± 2.9 pp, low confidence
41GPT-6.1 Sol53.9%estimated ± 3.1 pp, high confidence
42Ling 3.1 Flash53.8%estimated ± 3.1 pp, high confidence
43Gemini 3.1 Pro53.8%estimated ± 2.9 pp, low confidence
44Qwen3.7 Max53.5%measured
45GPT-5.6 Luna53.3%estimated ± 3.1 pp, high confidence
46Grok 4.653.1%estimated ± 2.9 pp, low confidence
47MiniMax M3.1 Flash53.1%estimated ± 3.1 pp, high confidence
48Gemini 3.5 Flash53.1%measured
49Ember-153.0%estimated ± 3.6 pp, medium confidence
50Gemini 3.6 Flash52.9%estimated ± 2.9 pp, low confidence
51GPT-5.2-Codex52.7%estimated ± 2.9 pp, low confidence
52Qwen3.8 Max52.5%estimated ± 2.9 pp, low confidence
53Claude Opus 4.852.3%estimated ± 2.9 pp, low confidence
54Kimi K2.652.2%measured
55Inkling 2 Small52.1%estimated ± 3.1 pp, high confidence
56Qwen3.8 Max Preview51.9%estimated ± 3.1 pp, high confidence
57DeepSeek V4 Pro51.7%estimated ± 2.9 pp, medium confidence
58DeepSeek V4.1 Flash 101751.7%estimated ± 3.1 pp, high confidence
59Grok 4.551.5%estimated ± 2.9 pp, medium confidence
60DeepSeek V4 Flash 073151.3%estimated ± 2.9 pp, medium confidence
61GPT-5.3 Codex51.3%estimated ± 2.9 pp, medium confidence
62Qwen3.7 Plus51.3%measured
63Composer 2.551.1%estimated ± 4.5 pp, medium confidence
64Kimi K351.1%estimated ± 2.9 pp, medium confidence
65MiMo-V2.6-Flash51.1%estimated ± 3.1 pp, high confidence
66GLM-5.251.0%estimated ± 3.1 pp, high confidence
67Qwen3.8-Flash-Next50.4%estimated ± 3.1 pp, high confidence
68Ornith-1.5-397B50.3%estimated ± 3.6 pp, medium confidence
69MiniMax M2.749.9%estimated ± 3.1 pp, high confidence
70GPT-5.149.7%estimated ± 2.9 pp, medium confidence
71Hy4 preview49.7%estimated ± 5.9 pp, medium confidence
72Gemini 3 Pro49.5%estimated ± 2.9 pp, medium confidence
73BTL-349.3%estimated ± 4.9 pp, high confidence
74Beam 1.0 501B-A23B49.2%measured
75Qwen3.6 Plus48.7%estimated ± 2.9 pp, medium confidence
76Inkling-Small48.7%measured
77Kimi K2.548.7%measured
78Ornith-1.0-397B48.7%estimated ± 3.6 pp, medium confidence
79Atria Dawn Preview48.6%estimated ± 5.9 pp, medium confidence
80Laguna S 2.148.5%estimated ± 5.9 pp, medium confidence
81GPT-548.5%estimated ± 2.9 pp, medium confidence
82GPT-5.6 Terra48.5%estimated ± 2.9 pp, medium confidence
83Muse Spark 1.148.5%estimated ± 2.9 pp, medium confidence
84Hy348.5%estimated ± 3.1 pp, high confidence
85Beam 1.0 501B-A23B FP848.3%measured
86Claude Opus 4.548.0%estimated ± 3.6 pp, high confidence
87Quasar 438B48.0%estimated ± 3.1 pp, high confidence
88Claude Opus 4.648.0%estimated ± 3.6 pp, high confidence
89Solar Pro 447.9%estimated ± 3.1 pp, high confidence
90Gemini 3 Flash47.9%estimated ± 2.9 pp, medium confidence
91GPT-5.1-Codex47.9%estimated ± 2.9 pp, medium confidence
92Nemotron 3.5 Super VL 120B A12B47.7%estimated ± 3.1 pp, high confidence
93Nemotron 3.5 Super VL 120B A12B NVFP447.7%estimated ± 3.1 pp, high confidence
94Inkling47.7%estimated ± 2.9 pp, medium confidence
95Mercury 347.7%estimated ± 7.6 pp, low confidence
96GPT-5.247.5%estimated ± 2.9 pp, medium confidence
97Grok 4.347.3%measured
98GPT-5.547.3%estimated ± 2.9 pp, medium confidence
99Inkling 2 Mini47.2%estimated ± 3.6 pp, high confidence
100Claude Sonnet 4.5 Thinking47.2%estimated ± 4.5 pp, medium confidence
101Ornith-1.5-35B-A3B47.2%estimated ± 3.6 pp, high confidence
102Qwen 3.6 Max (preview)47.0%measured
103Beam 1.0 501B-A23B NVFP446.9%measured
104Claude Opus 4.746.9%estimated ± 2.9 pp, medium confidence
105BTL-446.9%estimated ± 3.6 pp, high confidence
106dots3-note Preview46.9%estimated ± 3.6 pp, high confidence
107MiMo-V2-Pro46.7%estimated ± 3.6 pp, high confidence
108GLM-546.6%estimated ± 3.6 pp, high confidence
109Muse Spark46.4%estimated ± 3.6 pp, high confidence
110Claude Sonnet 4.546.3%estimated ± 3.6 pp, high confidence
111Gemini 2.5 Pro46.3%estimated ± 3.1 pp, high confidence
112Claude Opus 4.6 (Adaptive)46.1%estimated ± 2.9 pp, medium confidence
113Ornith-1.0-35B45.6%estimated ± 3.6 pp, high confidence
114Apodex 1.145.5%estimated ± 3.1 pp, high confidence
115Apodex 1.1 Mini45.5%estimated ± 3.1 pp, high confidence
116Gemma 4 31B45.5%estimated ± 3.1 pp, high confidence
117Grok 4.2045.3%estimated ± 2.9 pp, medium confidence
118MiMo-V2-Omni45.2%estimated ± 3.6 pp, high confidence
119Laguna M.145.1%estimated ± 3.6 pp, high confidence
120Claude 4.1 Opus45.1%estimated ± 3.6 pp, high confidence
121Step 3.5 Flash45.0%estimated ± 3.6 pp, high confidence
122Flint44.9%estimated ± 3.6 pp, high confidence
123GPT-5.444.9%estimated ± 2.9 pp, medium confidence
124Qwen3.5 397B44.8%estimated ± 3.1 pp, high confidence
125GPT-5.4 nano44.7%estimated ± 2.9 pp, medium confidence
126Qwen3.8-27B44.7%estimated ± 2.9 pp, medium confidence
127MAI-Thinking-144.6%estimated ± 3.6 pp, high confidence
128Nemotron 3 Ultra44.6%measured
129MiMo-V2-Flash44.6%estimated ± 3.6 pp, high confidence
130Claude Haiku 4.544.5%estimated ± 3.6 pp, high confidence
131o344.5%estimated ± 2.9 pp, low confidence
132Sage-A1.544.3%measured
133Jais-3-MoVA-36B-A4B44.3%estimated ± 3.6 pp, high confidence
134Claude 4 Sonnet44.2%estimated ± 3.6 pp, high confidence
135Ling 3.0 Flash VL44.2%estimated ± 3.1 pp, high confidence
136MAI-Code-1.1-Flash44.2%estimated ± 3.6 pp, high confidence
137Claude 4 Opus44.1%estimated ± 3.6 pp, high confidence
138Jais-3-32B44.1%estimated ± 3.6 pp, high confidence
139Qwen3.5-27B44.1%estimated ± 3.6 pp, high confidence
140Claude Opus 4.5 Thinking44.1%estimated ± 2.9 pp, low confidence
141Gemini 3.1 Flash-Lite44.0%estimated ± 4.5 pp, low confidence
142Step 3.7 Flash43.9%estimated ± 3.1 pp, high confidence
143GPT-5.1-Codex-Max43.9%estimated ± 2.9 pp, low confidence
144Muse Glimmer 30B43.6%measured
145Laguna XS 2.143.4%estimated ± 3.6 pp, high confidence
146Grok Code Fast 143.3%estimated ± 3.6 pp, high confidence
147Qwen3.5 Flash43.3%estimated ± 2.9 pp, low confidence
148Ornith-1.5-9B43.2%estimated ± 3.6 pp, high confidence
149Solar Open 243.1%estimated ± 3.6 pp, high confidence
150Grok 443.1%estimated ± 2.9 pp, low confidence
151Jais-3-375B-A23B43.0%estimated ± 3.6 pp, high confidence
152Jais-3-7B43.0%estimated ± 3.6 pp, high confidence
153Laguna XS.242.9%estimated ± 3.6 pp, high confidence
154Qwen3.6-27B42.9%estimated ± 3.1 pp, high confidence
155Qwen3-Coder 480B-A35B42.7%estimated ± 3.6 pp, high confidence
156Ornith-1.0-9B42.7%estimated ± 3.6 pp, high confidence
157Qwen3.5-35B-A3B42.6%estimated ± 3.6 pp, high confidence
158LongCat-Flash-Lite-Sparse42.1%estimated ± 3.6 pp, high confidence
159GPT-5.6 Sol41.9%estimated ± 2.9 pp, low confidence
160Claude Sonnet 541.5%estimated ± 2.9 pp, low confidence
161Gemini 3.5 Flash-Lite41.4%estimated ± 3.1 pp, high confidence
162Ling 3.0 Flash41.2%measured
163Hy3 Preview41.2%measured
164GLM-4.741.1%estimated ± 2.9 pp, low confidence
165MiniMax M341.1%estimated ± 2.9 pp, low confidence
166o4-mini41.1%estimated ± 2.9 pp, low confidence
167A.X K241.0%measured
168DeepSeek V3.141.0%estimated ± 3.6 pp, high confidence
169Claude Sonnet 4.640.9%estimated ± 2.9 pp, low confidence
170Kimi K2.7 Code40.9%estimated ± 2.9 pp, low confidence
171Kimi K240.9%estimated ± 3.6 pp, high confidence
172Trinity-Large-Preview40.7%estimated ± 3.1 pp, high confidence
173Trinity-Large-Thinking40.7%estimated ± 3.1 pp, high confidence
174o140.7%estimated ± 4.6 pp, high confidence
175Ling 3.0 Flash FP840.4%measured
176Mistral Medium 3.5 128B40.3%estimated ± 3.1 pp, high confidence
177Gemma 4 26B A4B40.1%estimated ± 3.1 pp, high confidence
178Nemotron 3.5 Bolt 74B-A9B39.9%estimated ± 3.1 pp, high confidence
179Qwen3.5-122B-A10B39.9%estimated ± 3.1 pp, high confidence
180GPT-5.4 mini39.7%estimated ± 2.9 pp, low confidence
181MiMo-V2.539.7%estimated ± 2.9 pp, low confidence
182GLM-5.139.5%estimated ± 2.9 pp, low confidence
183MiMo-V2.5-Pro39.5%estimated ± 2.9 pp, low confidence
184Ripple-65B Preview39.4%measured
185DeepSeek V3 032439.2%estimated ± 3.1 pp, high confidence
186Claude 3.7 Sonnet39.1%estimated ± 3.6 pp, high confidence
187GPT-OSS 20B39.1%estimated ± 3.1 pp, high confidence
188Mistral Small 439.0%estimated ± 3.1 pp, high confidence
189North Mini Code39.0%estimated ± 3.1 pp, high confidence
190Granite 4.2 30B38.8%measured
191GLM-4.638.8%estimated ± 2.9 pp, low confidence
192Command A+38.7%estimated ± 3.1 pp, high confidence
193Ternary Bonsai 2 27B38.4%estimated ± 3.6 pp, high confidence
194Nemotron 3 Super 100B38.2%estimated ± 3.6 pp, high confidence
195o1-preview38.2%estimated ± 4.6 pp, high confidence
196Mercury 2.538.0%measured
197Grok 4.1 Fast (Reasoning)38.0%estimated ± 2.9 pp, low confidence
198GLM-5.337.8%estimated ± 2.9 pp, low confidence
199GLM-5.3-Flash37.8%estimated ± 2.9 pp, low confidence
200GLM-4.7-Flash37.6%estimated ± 3.6 pp, high confidence
201K-EXAONE 2.037.4%measured
202Mistral Large 336.8%estimated ± 3.1 pp, high confidence
203Qwen3.6-35B-A3B36.8%estimated ± 3.1 pp, high confidence
204Spark-X2.6-4B36.8%measured
205Gemma 4 12B36.8%estimated ± 4.6 pp, high confidence
206DeepSeek-R1-052836.7%estimated ± 3.6 pp, high confidence
207Granite 4.2 8B36.1%measured
208DeepSeek V336.1%estimated ± 3.1 pp, high confidence
209MiniMax M1 80k35.9%estimated ± 3.6 pp, high confidence
210GPT-4.135.2%estimated ± 3.6 pp, high confidence
211ZAYA1-74B-Preview34.4%estimated ± 3.6 pp, high confidence
212GPT-OSS 120B34.3%estimated ± 3.1 pp, high confidence
213Gemini 1.5 Pro33.6%estimated ± 4.6 pp, high confidence
214LFM2.5-24B-A2B32.7%measured
215GPT-4 Turbo32.6%estimated ± 4.6 pp, high confidence
216K-Exaone32.3%estimated ± 3.6 pp, high confidence
217o3-mini32.3%estimated ± 3.6 pp, high confidence
218LLaDA2.2-flash32.3%estimated ± 3.6 pp, high confidence
219DeepSeek-R132.2%estimated ± 3.6 pp, high confidence
220Claude 3.5 Sonnet32.1%estimated ± 3.6 pp, high confidence
221Llama 4 Maverick32.1%estimated ± 3.1 pp, high confidence
222Nemotron 3 Nano Omni 30B A3B32.0%measured
223Claude 3 Opus31.8%estimated ± 4.6 pp, high confidence
224Nemotron 3.5 Lightning 30B A3B NVFP431.4%measured
225Mellum2.1-12B-A2.5B-Thinking31.0%estimated ± 3.6 pp, high confidence
226Devstral Small30.9%estimated ± 3.6 pp, high confidence
227Sarvam 105B29.9%estimated ± 3.6 pp, high confidence
228Mellum2-12B-A2.5B-Thinking29.6%estimated ± 4.9 pp, high confidence
229Agents-A1-4B29.6%measured
230Nemotron 3 Nano 30B29.5%estimated ± 4.6 pp, high confidence
231LFM3-1.7B28.5%measured
232Ultravox v0.6 Llama 3.3 70B28.4%estimated ± 4.6 pp, medium confidence
233Mistral Small 3.128.3%estimated ± 3.1 pp, high confidence
234GPT-4o mini28.2%estimated ± 4.6 pp, medium confidence
235GPT-4.1 nano28.1%estimated ± 4.6 pp, medium confidence
236Claude 3.5 Haiku27.4%estimated ± 3.6 pp, medium confidence
237Ling 2.6 Flash27.0%measured
238ZAYA1-8B26.5%estimated ± 4.9 pp, high confidence
239Gemma 4 E2B26.4%estimated ± 4.6 pp, medium confidence
240MiniCPM5-2B26.3%measured
241Solar Pro 326.1%estimated ± 3.1 pp, high confidence
242Gemma 4 E4B25.0%estimated ± 3.1 pp, medium confidence
243Ling 3.0 Tiny24.8%estimated ± 3.1 pp, medium confidence
244Granite 4.2 3B24.1%measured
245Gemma 3 27B23.9%estimated ± 3.1 pp, medium confidence
246Sarvam 30B23.4%estimated ± 3.6 pp, medium confidence
247Celeris-122.3%estimated ± 3.1 pp, medium confidence
248Llama 4 Scout22.0%estimated ± 3.1 pp, medium confidence
249Gemini 2.5 Flash-Lite22.0%estimated ± 3.6 pp, medium confidence
250Qwen3 235B 250717.7%estimated ± 4.9 pp, medium confidence
251Magistral Medium16.9%estimated ± 4.9 pp, medium confidence
252GPT-4.1 mini16.9%estimated ± 3.6 pp, medium confidence
253Magistral Small15.5%estimated ± 4.9 pp, medium confidence
254Gemini 2.0 Flash15.4%estimated ± 3.6 pp, medium confidence
255LFM2.5-2.6B15.3%estimated ± 3.1 pp, medium confidence
256Mellum2-12B-A2.5B-Instruct11.0%estimated ± 4.9 pp, medium confidence
257MiniCPM5-1B9.6%estimated ± 4.9 pp, medium confidence
258Gemini 2.0 Flash-Lite9.3%estimated ± 3.6 pp, medium confidence
259Mistral Medium 38.4%estimated ± 4.9 pp, medium confidence
260LLaDA2.2-mini7.7%estimated ± 4.9 pp, medium confidence
261Gemini 1.5 Flash7.2%estimated ± 3.6 pp, medium confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General