Benchmarks
ALE Bench
Models scored
87
Best score
2041.3
Metric
Performance
ALE Bench is an AI evaluation tracked by Epoch AI, with 87 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 April 2025 to June 2026. The best score recorded is 2041.3, measured as Performance. Models come from Anthropic, OpenAI, Google DeepMind, Alibaba, xAI, Moonshot, Z.ai (Zhipu AI), DeepSeek and others.
Full record
- Score metric
- Performance
- Models scored
- 87
- Best score recorded
- 2041.3
- Earliest model scored
- Apr 5, 2025
- Most recent model scored
- Jun 16, 2026
- Third-party reported
- Yes
Leaderboard
| 01 | claude-fable-5_high | Anthropic | 2041.3 |
| 02 | gpt-5.5_xhigh | OpenAI | 1943 |
| 03 | gpt-5.3-codex_xhigh | OpenAI | 1655.2 |
| 04 | gpt-5.4-2026-03-05_high | OpenAI | 1607 |
| 05 | gpt-5.5_medium | OpenAI | 1589.4 |
| 06 | claude-opus-4-8_high | Anthropic | 1563.8 |
| 07 | gpt-5.4-2026-03-05_medium | OpenAI | 1520.7 |
| 08 | claude-opus-4-8_none | Anthropic | 1411.8 |
| 09 | gemini-3-flash-preview | Google DeepMind | 1367.2 |
| 10 | claude-sonnet-4-6_medium | Anthropic | 1327.3 |
| 11 | claude-opus-4-7 | Anthropic | 1323 |
| 12 | gpt-5.2-codex | OpenAI | 1299.9 |
| 13 | gpt-5.2-2025-12-11_high | OpenAI | 1293.5 |
| 14 | gpt-5.2-2025-12-11_medium | OpenAI | 1249.8 |
| 15 | gpt-5.1-codex | OpenAI | 1244.9 |
| 16 | gpt-5.1-codex-max | OpenAI | 1208.8 |
| 17 | gpt-5.1-2025-11-13_high | OpenAI | 1192.2 |
| 18 | qwen3.7-max | Alibaba | 1189.4 |
| 19 | gpt-5.4-mini-2026-03-17_high | OpenAI | 1188.6 |
| 20 | gemini-3-pro-preview | Google DeepMind | 1176.8 |
| 21 | gpt-5-2025-08-07_high | OpenAI | 1162.5 |
| 22 | gemini-3.1-pro-preview | Google DeepMind | 1160.6 |
| 23 | grok-4-20 | xAI | 1150.3 |
| 24 | gpt-5.5_none | OpenAI | 1127.6 |
| 25 | kimi-k2.6 | Moonshot | 1092.7 |
| 26 | gpt-5.4-2026-03-05_none | OpenAI | 1086 |
| 27 | glm-5.2_high | Z.ai (Zhipu AI) | 1047 |
| 28 | claude-opus-4-5-20251101_16K | Anthropic | 1025.4 |
| 29 | glm-5.2_max | Z.ai (Zhipu AI) | 1010.2 |
| 30 | deepseek-v4-pro_high | DeepSeek | 1006.1 |
| 31 | gpt-5.4-nano-2026-03-17_high | OpenAI | 1004.5 |
| 32 | claude-opus-4-6 | Anthropic | 996.5 |
| 33 | grok-4-3 | xAI | 944.2 |
| 34 | o3-2025-04-16_high | OpenAI | 933.5 |
| 35 | gemma-4-26b-a4b | Google DeepMind | 927.2 |
| 36 | gemma-4-31b-it | Google DeepMind | 925.5 |
| 37 | gemini-3.5-flash_high | 911 | |
| 38 | mimo-v2.5-pro | Xiaomi Corp | 899.8 |
| 39 | glm-5.1 | Z.ai (Zhipu AI) | 887.1 |
| 40 | kimi-k2.7-code | Moonshot | 886.2 |
| 41 | o4-mini-2025-04-16_high | OpenAI | 826.2 |
| 42 | kimi-k2.5 | Moonshot | 821.6 |
| 43 | gpt-5-2025-08-07_minimal | OpenAI | 807.6 |
| 44 | DeepSeek-R1-0528 | DeepSeek | 804.1 |
| 45 | gpt-5-mini-2025-08-07_high | OpenAI | 799.8 |
| 46 | gemini-3.1-flash-lite | 797.7 | |
| 47 | claude-sonnet-4-5-20250929_32K | Anthropic | 796.1 |
| 48 | mercury-2 | Inception Labs | 785.6 |
| 49 | gemini-2.5-pro_32K | Google DeepMind | 785.5 |
| 50 | mimo-v2-pro | Xiaomi Corp | 785.2 |
| 51 | glm-5 | Z.ai (Zhipu AI) | 765.6 |
| 52 | mimo-v2-flash | Xiaomi Corp | 738 |
| 53 | gpt-5-nano-2025-08-07_high | OpenAI | 718.7 |
| 54 | deepseek-v4-flash_high | DeepSeek | 678.2 |
| 55 | claude-opus-4-1-20250805_16K | Anthropic | 674.8 |
| 56 | qwen3.6-plus | Alibaba | 670.1 |
| 57 | gemini-2.5-flash | Google DeepMind | 661.9 |
| 58 | claude-sonnet-4-20250514_32K | Anthropic | 655.4 |
| 59 | claude-haiku-4-5-20251001_32K | Anthropic | 653.5 |
| 60 | MiniMax-M3 | MiniMax | 640 |
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