List your company

Scientific Problem Solving on SciBench-107

62.5Atkins ScoreClaude-Opus-4.5
-2.514.37531.2548.125Feb 24, 2026

Evaluation results

MethodLinks
Claude-Opus-4.5Framework=Generative Models2026.0262.537.562.562.556.2537.542.869053.8562.5308058.3356.07
Gemini-3-ProFramework=Generative Models2026.0237.537.52537.534.3862.514.294030.7768.7570907546.73
Gemini-3-ProFramework=CodeAct2026.0237.537.5502537.512.528.575030.7775907077.7848.6
Qwen3-235B-InstructFramework=CodeAct2026.0237.550502540.622538.15038.4681.25909086.1155.14
Qwen3-235B-InstructFramework=EmbodiedAct2026.0237.537.562.52540.62033.3355.5632.19759010086.1152.34
Gemini-3-ProFramework=EmbodiedAct2026.0237.525502534.3812.552.385043.5993.75809088.8956.07
GPT-5-miniFramework=Generative Models2026.022550502537.537.519.055030.7731.25708055.5641.12
Qwen3-8B-VL-InstructFramework=CodeAct2026.022512.5012.512.5014.2944.4419.0950808066.6732.71
Gemini-3-FlashFramework=CodeAct2026.022537.5012.518.7512.514.292015.3950809069.4434.58
GPT-5-miniFramework=CodeAct2026.02252537.5021.8814.2938.155.5637.6981.25909086.1148.6
GPT-5.2Framework=CodeAct2026.02252537.52528.122538.15038.4675908080.5649.53
Claude-Sonnet-4.5Framework=EmbodiedAct2026.02252562.537.537.512.538.15035.9758010083.3352.34
GPT-5.2Framework=EmbodiedAct2026.022525502531.2512.547.625041.0387.5909088.8954.21
Claude-Opus-4.5Framework=EmbodiedAct2026.02255062.537.543.7512.552.386046.1587.59010091.6760.75
Qwen3-8B-VL-InstructFramework=Generative Models2026.0212.52512.5012.512.59.521010.256.25102011.1111.21
Gemini-3-FlashFramework=Generative Models2026.0212.5252512.518.7512.519.052017.9525407041.6726.17
Qwen3-235B-InstructFramework=Generative Models2026.0212.562.5250252514.293020.5218.75605038.8928.04
Claude-Sonnet-4.5Framework=Generative Models2026.0212.55062.52537.55023.814033.3331.25909063.8944.86
Claude-Sonnet-4.5Framework=CodeAct2026.0212.537.5502531.252523.815030.7737.5808061.1141.12
Claude-Opus-4.5Framework=CodeAct2026.0212.537.562.52534.3842.8633.3355.5640.9837.5808061.1144.86
Qwen3-8B-VL-InstructFramework=EmbodiedAct2026.0212.500259.3814.2914.2933.3319.1731.2580505026.17
Gemini-3-FlashFramework=EmbodiedAct2026.0212.5255037.531.2512.533.334030.7762.5909077.7846.73
GPT-5-miniFramework=EmbodiedAct2026.0212.537.537.52528.1214.2942.8655.5640.2668.75909080.5649.53
GPT-5.2Framework=Generative Models2026.02037.562.537.534.3837.523.812025.647570807544.86