benchgap
Knowledge & reasoning

BioMysteryBench (human-solvable) leaderboard

As of 2026-11-21, the highest measured score on BioMysteryBench (human-solvable) is 90.1% by Claude Opus 5. 203 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Opus 590.1%measured
2Claude Fable 5.289.5%estimated ± 0.6 pp, low confidence
3Muse Wonder Full89.5%estimated ± 0.6 pp, low confidence
4GPT-5.6 Sol89.5%estimated ± 0.6 pp, low confidence
5GPT-6.1 Sol89.5%estimated ± 0.6 pp, medium confidence
6GPT-6 Astra89.5%estimated ± 0.6 pp, medium confidence
7GPT-6 Sol89.5%estimated ± 0.6 pp, medium confidence
8GPT-5.5 Pro89.5%estimated ± 0.6 pp, medium confidence
9GPT-5.4 Pro89.5%estimated ± 0.6 pp, medium confidence
10GPT-5.6 Terra89.5%estimated ± 0.6 pp, medium confidence
11Claude Fable 5.189.5%estimated ± 0.6 pp, medium confidence
12Claude Fable 589.5%estimated ± 0.6 pp, medium confidence
13Muse Wonder89.5%estimated ± 0.6 pp, medium confidence
14GPT-6.1 Terra89.5%estimated ± 0.6 pp, medium confidence
15Gemini 4 Argon89.5%estimated ± 0.6 pp, medium confidence
16GPT-5.589.5%estimated ± 0.6 pp, medium confidence
17MiMo-V2.6-Pro89.5%estimated ± 0.6 pp, medium confidence
18Muse Spark 1.389.4%estimated ± 0.6 pp, medium confidence
19Kimi K3.1 Pro89.4%estimated ± 0.6 pp, medium confidence
20GPT-5.489.4%estimated ± 0.6 pp, medium confidence
21Kimi K389.4%estimated ± 0.6 pp, medium confidence
22MiniMax M3.1 Flash89.3%estimated ± 0.6 pp, medium confidence
23Claude Opus 5.589.3%measured
24Claude Opus 4.889.2%estimated ± 0.6 pp, medium confidence
25GLM-5.289.2%estimated ± 0.6 pp, medium confidence
26Step 5 Preview89.2%estimated ± 0.6 pp, medium confidence
27Claude Sonnet 5.589.2%measured
28GPT-5.6 Luna89.2%estimated ± 0.6 pp, medium confidence
29Sage-A1.589.1%estimated ± 0.6 pp, medium confidence
30GPT-6 Luna89.0%estimated ± 0.6 pp, medium confidence
31GLM-5.389.0%estimated ± 0.6 pp, medium confidence
32Claude Haiku 5.588.9%estimated ± 0.6 pp, medium confidence
33Gemini 3.8 Flash88.8%measured
34DeepSeek V4 Pro88.8%estimated ± 0.6 pp, medium confidence
35Ling 3.1 Flash88.8%estimated ± 0.6 pp, medium confidence
36Gemini 3.1 Pro88.7%estimated ± 0.6 pp, medium confidence
37Grok 4.788.7%estimated ± 0.6 pp, medium confidence
38Muse Spark 1.288.7%estimated ± 0.6 pp, medium confidence
39Qwen3.8 Max Preview88.7%estimated ± 0.6 pp, medium confidence
40Beam 1.0 501B-A23B FP888.6%estimated ± 0.6 pp, medium confidence
41Grok 4.688.5%estimated ± 0.6 pp, medium confidence
42Claude Sonnet 588.4%estimated ± 0.6 pp, medium confidence
43GPT-5.3 Codex88.4%estimated ± 0.6 pp, medium confidence
44Hy4 preview88.4%estimated ± 0.6 pp, medium confidence
45DeepSeek V4 Flash 073188.3%estimated ± 0.6 pp, medium confidence
46Beam 1.0 501B-A23B88.2%estimated ± 0.6 pp, medium confidence
47Grok 4.587.8%estimated ± 0.6 pp, medium confidence
48Beam 1.0 501B-A23B NVFP487.8%estimated ± 0.6 pp, medium confidence
49Muse Spark 1.187.6%estimated ± 0.6 pp, medium confidence
50Inkling 2 Small87.5%estimated ± 1.4 pp, low confidence
51Muse Spark Open87.5%estimated ± 0.6 pp, medium confidence
52Gemini 3.7 Flash87.1%measured
53DeepSeek V4.1 Flash 101787.1%estimated ± 0.6 pp, medium confidence
54Qwen3.7 Max86.4%estimated ± 0.6 pp, low confidence
55Gemini 3.5 Flash86.1%estimated ± 0.6 pp, low confidence
56Claude Opus 4.6 (Adaptive)85.7%estimated ± 0.6 pp, low confidence
57Claude Opus 4.785.1%estimated ± 0.6 pp, low confidence
58MiMo-V2.6-Flash85.1%estimated ± 0.6 pp, low confidence
59GPT-5.284.7%estimated ± 0.6 pp, low confidence
60Muse Spark84.4%estimated ± 0.6 pp, low confidence
61Qwen3.8-Flash-Next84.3%estimated ± 0.6 pp, low confidence
62Nemotron 3.5 Super VL 120B A12B84.0%estimated ± 0.6 pp, low confidence
63Gemini 3.6 Flash83.8%estimated ± 0.6 pp, low confidence
64Nemotron 3.5 Super VL 120B A12B NVFP483.8%estimated ± 0.6 pp, low confidence
65GPT-5.4 mini83.3%estimated ± 0.6 pp, low confidence
66Kimi K2.7 Code83.3%estimated ± 0.6 pp, low confidence
67Mistral Large 483.3%estimated ± 0.6 pp, low confidence
68Quasar 438B82.9%estimated ± 0.6 pp, low confidence
69GPT-5.4 nano82.8%estimated ± 0.6 pp, low confidence
70Gemini 3 Pro82.7%estimated ± 0.6 pp, low confidence
71Qwen3.7 Plus82.7%estimated ± 0.6 pp, low confidence
72A.X K282.6%estimated ± 0.6 pp, low confidence
73GPT-5.2-Codex82.5%estimated ± 0.6 pp, low confidence
74Inkling-Small82.3%estimated ± 0.6 pp, low confidence
75Grok 4.382.2%estimated ± 0.6 pp, low confidence
76Kimi K2.682.2%estimated ± 0.6 pp, low confidence
77Solar Pro 481.9%estimated ± 0.6 pp, low confidence
78GPT-581.8%estimated ± 0.6 pp, low confidence
79GPT-5.1-Codex81.8%estimated ± 0.6 pp, low confidence
80GPT-5.1-Codex-Max81.8%estimated ± 0.6 pp, low confidence
81Inkling81.8%estimated ± 0.6 pp, low confidence
82Qwen3.8-27B81.8%estimated ± 0.6 pp, low confidence
83GPT-5.181.8%estimated ± 0.6 pp, low confidence
84Hy381.8%estimated ± 0.6 pp, low confidence
85Hy3 Preview81.8%estimated ± 0.6 pp, low confidence
86Apodex 1.181.8%estimated ± 0.6 pp, low confidence
87Apodex 1.1 Mini81.8%estimated ± 0.6 pp, low confidence
88Claude Opus 4.5 Thinking81.8%estimated ± 0.6 pp, low confidence
89GLM-5.181.8%estimated ± 0.6 pp, low confidence
90Nemotron 3.5 Bolt 74B-A9B81.8%estimated ± 0.6 pp, low confidence
91MiMo-V2.5-Pro81.7%estimated ± 0.6 pp, low confidence
92MiniMax M381.7%estimated ± 0.6 pp, low confidence
93Qwen 3.6 Max (preview)81.7%estimated ± 0.6 pp, low confidence
94Kimi K2.581.7%estimated ± 0.6 pp, low confidence
95Nemotron 3 Ultra81.7%estimated ± 0.6 pp, low confidence
96Grok 4.1 Fast (Reasoning)81.7%estimated ± 0.6 pp, low confidence
97Grok 4 Fast (Reasoning)81.7%estimated ± 0.6 pp, low confidence
98Qwen3.6 Plus81.7%estimated ± 0.6 pp, low confidence
99Claude Opus 4.681.7%estimated ± 0.6 pp, low confidence
100Gemini 2.5 Pro81.7%estimated ± 0.6 pp, low confidence
101Muse Glimmer 30B81.7%estimated ± 0.6 pp, low confidence
102Step 3.7 Flash81.7%estimated ± 0.6 pp, low confidence
103GLM-581.7%estimated ± 0.6 pp, low confidence
104Grok 481.7%estimated ± 0.6 pp, low confidence
105Ling 3.0 Flash VL81.7%estimated ± 0.6 pp, low confidence
106GLM-4.781.7%estimated ± 0.6 pp, low confidence
107Ling 3.0 Flash81.7%estimated ± 0.6 pp, low confidence
108Qwen3.5 397B81.7%estimated ± 0.6 pp, low confidence
109DeepSeek-R181.7%estimated ± 0.6 pp, low confidence
110Gemini 2.5 Flash81.7%estimated ± 0.6 pp, low confidence
111Gemini 3 Flash81.7%estimated ± 0.6 pp, low confidence
112Gemma 4 31B81.7%estimated ± 0.6 pp, low confidence
113GPT-OSS 20B81.7%estimated ± 0.6 pp, low confidence
114Claude 4 Sonnet81.7%estimated ± 0.6 pp, low confidence
115GPT-OSS 120B81.7%estimated ± 0.6 pp, low confidence
116K-Exaone81.7%estimated ± 0.6 pp, low confidence
117MiMo-V2-Omni81.7%estimated ± 0.6 pp, low confidence
118o381.7%estimated ± 0.6 pp, low confidence
119Qwen3.6-27B81.7%estimated ± 0.6 pp, low confidence
120Claude 3.7 Sonnet Thinking81.7%estimated ± 0.6 pp, low confidence
121Claude Sonnet 4.681.7%estimated ± 0.6 pp, low confidence
122DeepSeek V3.281.7%estimated ± 0.6 pp, low confidence
123K-EXAONE 2.081.7%estimated ± 0.6 pp, low confidence
124Qwen3.5-27B81.7%estimated ± 0.6 pp, low confidence
125Qwen3.5-35B-A3B81.7%estimated ± 0.6 pp, low confidence
126Trinity-Large-Preview81.7%estimated ± 0.6 pp, low confidence
127Trinity-Large-Thinking81.7%estimated ± 0.6 pp, low confidence
128Gemma 4 E4B81.7%estimated ± 0.6 pp, low confidence
129GLM-5V-Turbo81.7%estimated ± 0.6 pp, low confidence
130MiniMax M2.781.7%estimated ± 0.6 pp, low confidence
131o4-mini81.7%estimated ± 0.6 pp, low confidence
132Qwen3.5-122B-A10B81.7%estimated ± 0.6 pp, low confidence
133Claude Opus 4.581.7%estimated ± 0.6 pp, low confidence
134Command A+81.7%estimated ± 0.6 pp, low confidence
135GLM-5-Turbo81.7%estimated ± 0.6 pp, low confidence
136Granite 4.2 30B81.7%estimated ± 0.6 pp, low confidence
137Granite 4.2 8B81.7%estimated ± 0.6 pp, low confidence
138Magistral Medium81.7%estimated ± 0.6 pp, low confidence
139MiMo-V2-Pro81.7%estimated ± 0.6 pp, low confidence
140MiniCPM5-2B81.7%estimated ± 0.6 pp, low confidence
141Mistral Small 481.7%estimated ± 0.6 pp, low confidence
142North Mini Code81.7%estimated ± 0.6 pp, low confidence
143o181.7%estimated ± 0.6 pp, low confidence
144Qwen3.6-35B-A3B81.7%estimated ± 0.6 pp, low confidence
145Sarvam 30B81.7%estimated ± 0.6 pp, low confidence
146Celeris-181.7%estimated ± 0.6 pp, low confidence
147Claude 3.5 Haiku81.7%estimated ± 0.6 pp, low confidence
148Claude 3.7 Sonnet81.7%estimated ± 0.6 pp, low confidence
149Claude 3 Haiku81.7%estimated ± 0.6 pp, low confidence
150DeepSeek V381.7%estimated ± 0.6 pp, low confidence
151DeepSeek V3 032481.7%estimated ± 0.6 pp, low confidence
152DeepSeek V3.181.7%estimated ± 0.6 pp, low confidence
153Devstral Small81.7%estimated ± 0.6 pp, low confidence
154Exaone 4.0 1.2B81.7%estimated ± 0.6 pp, low confidence
155Exaone 4.0 32B81.7%estimated ± 0.6 pp, low confidence
156Gemini 2.0 Flash81.7%estimated ± 0.6 pp, low confidence
157Gemini 2.5 Flash-Lite81.7%estimated ± 0.6 pp, low confidence
158Gemini 3.5 Flash-Lite81.7%estimated ± 0.6 pp, low confidence
159Gemma 3 27B81.7%estimated ± 0.6 pp, low confidence
160Gemma 4 12B81.7%estimated ± 0.6 pp, low confidence
161Gemma 4 26B A4B81.7%estimated ± 0.6 pp, low confidence
162Gemma 4 E2B81.7%estimated ± 0.6 pp, low confidence
163GLM-4.5-Air81.7%estimated ± 0.6 pp, low confidence
164GLM-4.681.7%estimated ± 0.6 pp, low confidence
165GPT-4.181.7%estimated ± 0.6 pp, low confidence
166GPT-4.1 mini81.7%estimated ± 0.6 pp, low confidence
167GPT-4.1 nano81.7%estimated ± 0.6 pp, low confidence
168GPT-4o81.7%estimated ± 0.6 pp, low confidence
169Granite-4.0-350M81.7%estimated ± 0.6 pp, low confidence
170Granite-4.0-H-1B81.7%estimated ± 0.6 pp, low confidence
171Granite-4.0-H-350M81.7%estimated ± 0.6 pp, low confidence
172Granite 4.2 3B81.7%estimated ± 0.6 pp, low confidence
173Grok 4.1 Fast81.7%estimated ± 0.6 pp, low confidence
174Grok Code Fast 181.7%estimated ± 0.6 pp, low confidence
175Kimi K281.7%estimated ± 0.6 pp, low confidence
176LFM2.5-2.6B81.7%estimated ± 0.6 pp, low confidence
177LFM2.5-8B-A1B81.7%estimated ± 0.6 pp, low confidence
178LFM2.5-VL-1.6B-Extract81.7%estimated ± 0.6 pp, low confidence
179Ling 2.6 Flash81.7%estimated ± 0.6 pp, low confidence
180Ling 3.0 Tiny81.7%estimated ± 0.6 pp, low confidence
181Llama 3.1 405B81.7%estimated ± 0.6 pp, low confidence
182Llama 3.1 70B81.7%estimated ± 0.6 pp, low confidence
183Llama 3.3 70B81.7%estimated ± 0.6 pp, low confidence
184Llama 4 Maverick81.7%estimated ± 0.6 pp, low confidence
185Llama 4 Scout81.7%estimated ± 0.6 pp, low confidence
186Magistral Small81.7%estimated ± 0.6 pp, low confidence
187Mercury 2.581.7%estimated ± 0.6 pp, low confidence
188MiMo-V2-Flash81.7%estimated ± 0.6 pp, low confidence
189Mistral Large 281.7%estimated ± 0.6 pp, low confidence
190Mistral Large 381.7%estimated ± 0.6 pp, low confidence
191Mistral Medium 381.7%estimated ± 0.6 pp, low confidence
192Mistral Medium 3.5 128B81.7%estimated ± 0.6 pp, low confidence
193Mistral Small 3.181.7%estimated ± 0.6 pp, low confidence
194Nemotron 3.5 Lightning 30B A3B NVFP481.7%estimated ± 0.6 pp, low confidence
195Nemotron 3 Nano 30B81.7%estimated ± 0.6 pp, low confidence
196Nemotron 3 Nano Omni 30B A3B81.7%estimated ± 0.6 pp, low confidence
197Nemotron Ultra 253B81.7%estimated ± 0.6 pp, low confidence
198Nova Pro81.7%estimated ± 0.6 pp, low confidence
199Phi-481.7%estimated ± 0.6 pp, low confidence
200Phi-4-reasoning81.7%estimated ± 0.6 pp, low confidence
201Qwen3.5-4B81.7%estimated ± 0.6 pp, low confidence
202Qwen3 Max81.7%estimated ± 0.6 pp, low confidence
203Qwen3-Omni-30B-A3B-Instruct81.7%estimated ± 0.6 pp, low confidence
204Qwen3-Omni-30B-A3B-Thinking81.7%estimated ± 0.6 pp, low confidence
205Sarvam 105B81.7%estimated ± 0.6 pp, low confidence
206Solar Pro 281.7%estimated ± 0.6 pp, low confidence
207Solar Pro 381.7%estimated ± 0.6 pp, low confidence
208Ultravox v0.6 Llama 3.3 70B81.7%estimated ± 0.6 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General