Inference
EN
Vals AI

BioMysteryBench

Anthropic's benchmark of whether AI agents can solve real-world bioinformatics mysteries from raw data

ModelsModels with a result.30
Top resultBest result on this benchmark.83.7%Claude Fable 5.2
Top-3 spreadPoints from first to third.4.1 pts
UpdatedDate the source changed the results.9 Nov 2026

Result and cost

29 of 30 results

Results

30 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01Claude Fable 5.2Terminus 283.7% ±0.7$1.684 min
02Claude Sonnet 5.5Terminus 281.1% ±0.6$2.9518 min
03GPT-6.1 SolTerminus 279.6% ±0.7$0.3729 min
04Claude Opus 5.5Terminus 279.3% ±1.0$2.9512 min
04GPT-6 AstraTerminus 279.3% ±1.0$1.725 min
04Claude Opus 5Terminus 279.3% ±2.1$3.2937 min
07Gemini 4 ArgonTerminus 276.3% ±0.4$5.7140 min
08GPT-6 SolTerminus 274.8% ±2.4$0.6625 min
09Muse Wonder FullTerminus 273.3% ±0.6$3.4726 min
10Kimi K3Terminus 272.6% ±1.6$1.065 min
11Grok 4.6Terminus 272.2%$1.2314 min
12GPT-5.6 SolTerminus 271.1% ±0.6$1.9318 min
13Claude Haiku 5.5Terminus 270.4% ±1.3$0.51313 min
14Inkling 2 SmallTerminus 270.0% ±1.3$0.2698 min
15Grok 4.7Terminus 269.3% ±1.3$2.4912 min
15Hy4 previewTerminus 269.3% ±2.6$0.36224 min
15MiMo-V2.6-FlashTerminus 269.3% ±1.0$0.0528 min
18DeepSeek V4.1 Flash 1017Terminus 267.8% ±1.1$0.23315 min
19Step 5 PreviewTerminus 267.4% ±0.4$0.49522 min
20Muse WonderTerminus 267.0% ±0.4$3.2318 min
20Mistral Large 4Terminus 267.0% ±2.7$1.0526 min
20Ember-1Terminus 267.0% ±3.0$0.9079 min
23Muse Spark 1.2Terminus 264.8% ±1.3$0.99524 min
24DeepSeek V4 Flash 0731Terminus 264.4% ±1.3$0.03927 min
25Gemini 3.8 FlashTerminus 262.2% ±1.7$1.555 min
26GPT-6 LunaTerminus 261.5% ±2.6$0.0539 min
26GPT-5.6 LunaTerminus 261.5% ±0.7$0.08919 min
28Gemini 3.1 Pro PreviewTerminus 260.0% ±2.9$0.7134 min
29Gemini 3.6 FlashTerminus 258.5% ±1.6$1.2513 min
30Beam 1.0Terminus 242.6% ±1.6—7 min