SciCode
46 models
SciCode measures an AI's ability to generate executable scientific code that solves real research problems — including data analysis, simulations, and modeling — based on natural language descriptions from scientific papers, testing both coding proficiency and domain-specific scientific understanding.
Top 5 Models Performance
| Anthropic: claude-sonnet-5.5 | ######################################## | 61 |
| sakana: fugu | ####################################### | 60.1 |
| Anthropic: Claude Fable 5 | ####################################### | 60 |
| Moonshot AI: kimi-k3 | ####################################### | 59 |
| Google: Gemini 3.1 Pro Preview | ####################################### | 59 |