SciCode leaderboard
35 models ranked, highest score first.
| # | Model | Company | Score |
|---|---|---|---|
| 1 | Claude Fable 5 | Anthropic | 60.2% |
| 2 | Claude Opus 4.8 | Anthropic | 53.5% |
| 3 | GPT-5.3 Codex | OpenAI | 53.2% |
| 4 | GPT-5.2 | OpenAI | 52.1% |
| 5 | Claude Opus 4.5 | Anthropic | 49.5% |
| 6 | Qwen3.7 Max | Alibaba | 48.8% |
| 7 | Gemini 3.5 Flash | 48.8% | |
| 8 | Qwen3.7 Plus | Alibaba | 45.5% |
| 9 | MiniMax-M3 | MiniMax | 45.4% |
| 10 | Claude Sonnet 4.6 | Anthropic | 44.1% |
| 11 | GPT-5.1 | OpenAI | 43.3% |
| 12 | Ring-2.6-1T | InclusionAI | 42.4% |
| 13 | Step 3.7 Flash | StepFun | 40.0% |
| 14 | Gemini 2.5 Flash | 39.4% | |
| 15 | DeepSeek V3.2 | DeepSeek | 38.9% |
| 16 | GPT-5 | OpenAI | 38.8% |
| 17 | North Mini Code | Cohere | 38.2% |
| 18 | Gemma 4 12B | 38.2% | |
| 19 | Mistral Large 3 | Mistral | 36.2% |
| 20 | DeepSeek-V3 | DeepSeek | 35.4% |
| 21 | Gemini 2.0 Flash | 34.0% | |
| 22 | HyperNova 60B 2605 | Multiverse Computing | 33.0% |
| 23 | GPT-4 Turbo | OpenAI | 31.9% |
| 24 | Claude 3.5 Sonnet | Anthropic | 31.6% |
| 25 | JT-35B-Flash | China Mobile | 29.1% |
| 26 | Grok-2 | xAI | 28.5% |
| 27 | Gemini 1.5 Pro | 27.4% | |
| 28 | Claude 3 Opus | Anthropic | 23.3% |
| 29 | Claude 3 Sonnet | Anthropic | 22.9% |
| 30 | Mistral Large | Mistral | 20.8% |
| 31 | Claude 3 Haiku | Anthropic | 18.6% |
| 32 | Claude 2.1 | Anthropic | 18.4% |
| 33 | LFM2.5-8B-A1B | Liquid AI | 7.8% |
| 34 | MiniCPM5-1B | OpenBMB | 4.4% |
| 35 | MiniCPM-V 4.6 1.3B | OpenBMB | 2.1% |