Benchmarks
Lech Mazur Writing
Models scored
49
Best score
8.6
Metric
Mean score
Lech Mazur Writing is an AI evaluation tracked by Epoch AI, with 49 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 August 2025. The best score recorded is 8.6, measured as Mean score. Models come from OpenAI, Alibaba, DeepSeek, Anthropic, Google DeepMind, xAI, Google DeepMind,Google, Mistral AI and others.
Full record
- Score metric
- Mean score
- Models scored
- 49
- Best score recorded
- 8.6
- Earliest model scored
- Jul 18, 2024
- Most recent model scored
- Aug 7, 2025
- Third-party reported
- Yes
Leaderboard
| 01 | gpt-5-2025-08-07_medium | OpenAI | 8.6 |
| 02 | Kimi-K2-Instruct | Moonshot | 8.56 |
| 03 | claude-opus-4-1-20250805 | Anthropic | 8.47 |
| 04 | claude-opus-4-1-20250805_16K | Anthropic | 8.45 |
| 05 | o3-pro-2025-06-10_medium | OpenAI | 8.44 |
| 06 | o3-2025-04-16_medium | OpenAI | 8.39 |
| 07 | gemini-2.5-pro | Google DeepMind | 8.38 |
| 08 | claude-opus-4-20250514_16K | Anthropic | 8.36 |
| 09 | claude-opus-4-20250514 | Anthropic | 8.31 |
| 10 | gpt-5-mini-2025-08-07_medium | OpenAI | 8.31 |
| 11 | qwen3-235b-a22b | Alibaba | 8.3 |
| 12 | DeepSeek-R1 | DeepSeek | 8.3 |
| 13 | Qwen3-235B-A22B-Thinking-2507 | Alibaba | 8.24 |
| 14 | DeepSeek-R1-0528 | DeepSeek | 8.19 |
| 15 | gpt-4o-2024-11-20 | OpenAI | 8.18 |
| 16 | claude-sonnet-4-20250514_16K | Anthropic | 8.14 |
| 17 | claude-3-7-sonnet-20250219_16K | Anthropic | 8.11 |
| 18 | claude-sonnet-4-20250514 | Anthropic | 8.09 |
| 19 | gemini-2.5-pro-preview-05-06 | Google DeepMind | 8.09 |
| 20 | gemini-2.5-pro-exp-03-25 | Google DeepMind | 8.05 |
| 21 | claude-3-5-sonnet-20241022 | Anthropic | 8.03 |
| 22 | QwQ-32B (16K thinking) | Alibaba | 8.02 |
| 23 | gemma-3-27b-it | Google DeepMind | 7.99 |
| 24 | claude-3-7-sonnet-20250219 | Anthropic | 7.94 |
| 25 | mistral-medium-2505 | Mistral AI | 7.73 |
| 26 | gpt-oss-120b | OpenAI | 7.71 |
| 27 | DeepSeek-V3-0324 | DeepSeek | 7.7 |
| 28 | grok-4-0709 | xAI | 7.69 |
| 29 | gemini-2.5-flash-preview-04-17 (24K thinking) | Google DeepMind | 7.65 |
| 30 | grok-3-beta | xAI | 7.64 |
| 31 | gpt-4.5-preview-2025-02-27 | OpenAI | 7.56 |
| 32 | Qwen3-30B-A3B | Alibaba | 7.53 |
| 33 | o4-mini-2025-04-16_medium | OpenAI | 7.5 |
| 34 | gemini-2.0-flash-thinking-exp-01-21 | Google DeepMind,Google | 7.38 |
| 35 | grok-3-mini-beta_low | xAI | 7.35 |
| 36 | claude-3-5-haiku-20241022 | Anthropic | 7.35 |
| 37 | glm-4.5 | Z.ai (Zhipu AI),Tsinghua University | 7.34 |
| 38 | qwen2.5-max | Alibaba | 7.29 |
| 39 | gemini-2.0-flash-exp | Google DeepMind,Google | 7.15 |
| 40 | o1-2024-12-17_medium | OpenAI | 7.02 |
| 41 | mistral-large-2407 | Mistral AI | 6.9 |
| 42 | gpt-4o-mini-2024-07-18 | OpenAI | 6.72 |
| 43 | o1-mini-2024-09-12_medium | OpenAI | 6.49 |
| 44 | grok-2-1212 | xAI | 6.36 |
| 45 | phi-4 | Microsoft Research | 6.26 |
| 46 | Llama-4-Maverick-17B-128E-Instruct | Meta AI | 6.2 |
| 47 | o3-mini-2025-01-31_high | OpenAI | 6.17 |
| 48 | o3-mini-2025-01-31_medium | OpenAI | 6.15 |
| 49 | amazon.nova-pro-v1:0 | Amazon | 6.05 |
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