Benchmarks
Apex Agents
Models scored
50
Best score
0.5
Metric
Pass@1 score
Apex Agents is an AI evaluation tracked by Epoch AI, with 50 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 November 2024 to July 2026. The best score recorded is 0.5, measured as Pass@1 score. Models come from Google, Anthropic, OpenAI, Z.ai (Zhipu AI), Google DeepMind, Moonshot, xAI, Alibaba and others.
Full record
- Score metric
- Pass@1 score
- Models scored
- 50
- Best score recorded
- 0.5
- Earliest model scored
- Nov 20, 2024
- Most recent model scored
- Jul 24, 2026
- Third-party reported
- Yes
Leaderboard
| 01 | gemini-3.5-flash_high | 0.5 | |
| 02 | gemini-3.5-flash_unknown | 0.5 | |
| 03 | claude-fable-5 | Anthropic | 0.45 |
| 04 | claude-opus-5_max | Anthropic | 0.43 |
| 05 | claude-opus-4-8_max | Anthropic | 0.42 |
| 06 | gpt-5.5_xhigh | OpenAI | 0.38 |
| 07 | gpt-5.4-2026-03-05_xhigh | OpenAI | 0.36 |
| 08 | glm-5.2_unknown | Z.ai (Zhipu AI) | 0.36 |
| 09 | gpt-5.2-2025-12-11_xhigh | OpenAI | 0.34 |
| 10 | claude-opus-4-7_max | Anthropic | 0.34 |
| 11 | gemini-3.1-pro-preview | Google DeepMind | 0.34 |
| 12 | gpt-5.3-codex_high | OpenAI | 0.32 |
| 13 | claude-opus-4-6_max | Anthropic | 0.32 |
| 14 | claude-opus-4-6_unknown | Anthropic | 0.3 |
| 15 | gpt-5.2-codex | OpenAI | 0.28 |
| 16 | gpt-5.4-mini-2026-03-17_xhigh | OpenAI | 0.25 |
| 17 | gemini-3-flash-preview | Google DeepMind | 0.24 |
| 18 | claude-sonnet-4-6_high | Anthropic | 0.24 |
| 19 | gpt-5.2-2025-12-11_high | OpenAI | 0.23 |
| 20 | gpt-5.1-codex | OpenAI | 0.21 |
| 21 | gpt-5-codex | OpenAI | 0.2 |
| 22 | kimi-k2.6 | Moonshot | 0.19 |
| 23 | claude-opus-4-5-20251101_unknown | Anthropic | 0.18 |
| 24 | gemini-3-pro-preview | Google DeepMind | 0.18 |
| 25 | gpt-5-2025-08-07_high | OpenAI | 0.18 |
| 26 | gpt-5.1-2025-11-13_high | OpenAI | 0.18 |
| 27 | glm-5 | Z.ai (Zhipu AI) | 0.17 |
| 28 | o3-2025-04-16_high | OpenAI | 0.17 |
| 29 | gpt-5.4-nano-2026-03-17_xhigh | OpenAI | 0.17 |
| 30 | grok-4-0709 | xAI | 0.15 |
| 31 | kimi-k2.5 | Moonshot | 0.14 |
| 32 | qwen3.5-plus | Alibaba | 0.14 |
| 33 | gemini-3.1-flash-lite | 0.13 | |
| 34 | grok-4-1-fast-reasoning | xAI | 0.13 |
| 35 | nemotron-3-ultra | NVIDIA | 0.12 |
| 36 | claude-sonnet-4-20250514 | Anthropic | 0.09 |
| 37 | claude-sonnet-4-20250514_unknown | Anthropic | 0.09 |
| 38 | claude-haiku-4-5-20251001_32K | Anthropic | 0.09 |
| 39 | claude-haiku-4-5-20251001_unknown | Anthropic | 0.09 |
| 40 | deepseek/deepseek-v3.2 | DeepSeek | 0.07 |
| 41 | gemini-2.5-pro | Google DeepMind | 0.07 |
| 42 | MiniMax-M2.5 | MiniMax | 0.06 |
| 43 | gpt-oss-120b_high | OpenAI | 0.05 |
| 44 | kimi-k2-thinking | Moonshot | 0.04 |
| 45 | zai-org/glm-4.7 | Z.ai (Zhipu AI) | 0.03 |
| 46 | glm-4.6 | Z.ai (Zhipu AI),Tsinghua University | 0.03 |
| 47 | grok-3 | xAI | 0.02 |
| 48 | gemini-2.5-flash | Google DeepMind | 0.02 |
| 49 | o1-2024-12-17_high | OpenAI | 0.01 |
| 50 | gpt-4o-2024-11-20 | OpenAI | 0.01 |
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