Benchmarks
SciCode
Models scored
75
Best score
0.6
Metric
Score
SciCode is an AI evaluation tracked by Epoch AI, with 75 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.
Scored models span July 2024 to July 2026. The best score recorded is 0.6, measured as Score. Models come from Anthropic, Google DeepMind, OpenAI, Moonshot, Google, Meta AI, Z.ai (Zhipu AI), Xiaomi Corp and others.
Full record
- Score metric
- Score
- Models scored
- 75
- Best score recorded
- 0.6
- Earliest model scored
- Jul 23, 2024
- Most recent model scored
- Jul 24, 2026
- Third-party reported
- Yes
Leaderboard
| 01 | claude-fable-5_max | Anthropic | 0.6 |
| 02 | gemini-3.1-pro-preview | Google DeepMind | 0.59 |
| 03 | gpt-5.6-sol_high | OpenAI | 0.57 |
| 04 | gpt-5.4-2026-03-05_xhigh | OpenAI | 0.57 |
| 05 | gpt-5.6-sol_medium | OpenAI | 0.56 |
| 06 | gpt-5.6-sol_max | OpenAI | 0.56 |
| 07 | gpt-5.5_xhigh | OpenAI | 0.56 |
| 08 | gpt-5.6-sol_xhigh | OpenAI | 0.56 |
| 09 | gpt-5.5_high | OpenAI | 0.56 |
| 10 | claude-opus-5_max | Anthropic | 0.56 |
| 11 | gpt-5.6-sol_low | OpenAI | 0.55 |
| 12 | claude-opus-4-7_max | Anthropic | 0.55 |
| 13 | grok-4.5_high | xAI | 0.54 |
| 14 | gpt-5.6-terra_max | OpenAI | 0.54 |
| 15 | claude-sonnet-5_max | Anthropic | 0.54 |
| 16 | kimi-k2.6 | Moonshot | 0.53 |
| 17 | claude-opus-4-8_max | Anthropic | 0.53 |
| 18 | gpt-5.5_medium | OpenAI | 0.53 |
| 19 | gemini-3.5-flash_high | 0.53 | |
| 20 | gpt-5.6-luna_max | OpenAI | 0.53 |
| 21 | gpt-5.5_low | OpenAI | 0.52 |
| 22 | gpt-5.6-terra_xhigh | OpenAI | 0.52 |
| 23 | muse-spark | Meta AI | 0.52 |
| 24 | gpt-5.6-luna_high | OpenAI | 0.51 |
| 25 | glm-5.2_max | Z.ai (Zhipu AI) | 0.5 |
| 26 | mimo-v2.5-pro | Xiaomi Corp | 0.5 |
| 27 | gpt-5.6-terra_high | OpenAI | 0.5 |
| 28 | deepseek-v4-pro_max | DeepSeek | 0.5 |
| 29 | gpt-5.6-luna_xhigh | OpenAI | 0.5 |
| 30 | gpt-5.4-mini-2026-03-17_xhigh | OpenAI | 0.5 |
| 31 | gpt-5.6-terra_medium | OpenAI | 0.5 |
| 32 | gpt-5.6-terra_low | OpenAI | 0.49 |
| 33 | qwen3.7-max | Alibaba | 0.49 |
| 34 | gpt-5.5_none | OpenAI | 0.49 |
| 35 | kimi-k2.7-code | Moonshot | 0.47 |
| 36 | grok-4.3_high | xAI | 0.47 |
| 37 | MiniMax-M2.7 | MiniMax | 0.47 |
| 38 | gpt-5.4-nano-2026-03-17_xhigh | OpenAI | 0.47 |
| 39 | claude-sonnet-4-6_max | Anthropic | 0.47 |
| 40 | gpt-5.6-luna_medium | OpenAI | 0.46 |
| 41 | gpt-5.6-luna_low | OpenAI | 0.46 |
| 42 | MiniMax-M3 | MiniMax | 0.45 |
| 43 | deepseek-v4-flash_max | DeepSeek | 0.45 |
| 44 | glm-5.1 | Z.ai (Zhipu AI) | 0.44 |
| 45 | gemma-4-31b-it | Google DeepMind | 0.43 |
| 46 | claude-haiku-4-5-20251001_unknown | Anthropic | 0.43 |
| 47 | gpt-5.1-2025-11-13_unknown | OpenAI | 0.43 |
| 48 | gpt-5-2025-08-07_unknown | OpenAI | 0.43 |
| 49 | gemini-2.5-pro | Google DeepMind | 0.43 |
| 50 | gemini-3.1-flash-lite | 0.42 | |
| 51 | qwen3.6-plus | Alibaba | 0.41 |
| 52 | gpt-4.1-mini-2025-04-14 | OpenAI | 0.4 |
| 53 | gemma-4-26b-a4b | Google DeepMind | 0.4 |
| 54 | claude-sonnet-4-20250514_unknown | Anthropic | 0.4 |
| 55 | nemotron-3-ultra | NVIDIA | 0.4 |
More benchmarks
SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.
← All benchmarks