Long-horizon Task Solving on PINCHBENCH 23 tasks original
93.3Best Success RateClaude-Opus-4.6
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Claude-Opus-4.6Source type=Closed-source models2026.06 | 93.3 | 81.6 | |
| GPT-5.4Source type=Closed-source models2026.06 | 90.5 | 79.4 | |
| MiniMax-M2.7Source type=Open-source / open-weight models2026.06 | 89.8 | 82.8 | |
| Claude-Haiku-4.5Source type=Closed-source models2026.06 | 89.5 | 77.4 | |
| MiMo-V2-FlashSource type=Open-source / open-weight models2026.06 | 88.8 | 69.7 | |
| MiMo-V2-ProSource type=Closed-source models2026.06 | 87.4 | 80.4 | |
| Gemini-3.1-Pro-PreviewSource type=Closed-source models2026.06 | 86.7 | 77.5 | |
| OpenClaw-Skill 9BSource type=Open-source / open-weight models2026.06 | 72.8 | 58.9 | |
| OpenClaw-Skill 4BSource type=Open-source / open-weight models2026.06 | 71.5 | 56.4 | |
| Qwen3.5-4BSource type=Open-source / open-weight models2026.06 | 71 | 55.7 | |
| Qwen3.5-9BSource type=Open-source / open-weight models2026.06 | 67.5 | 53.8 | |
| GPT-OSS-120BSource type=Open-source / open-weight models2026.06 | 67.1 | 52 | |
| GPT-OSS-20BSource type=Open-source / open-weight models2026.06 | 66 | 50.3 | |
| OpenClaw-Skill-Qwen3 8BSource type=Open-source / open-weight models2026.06 | 64.9 | 49.2 | |
| OpenClaw-Skill-Qwen3-4BSource type=Open-source / open-weight models2026.06 | 64.5 | 47.9 | |
| Qwen3.6-PlusSource type=Open-source / open-weight models2026.06 | 63.9 | 63.3 | |
| Qwen3-8BSource type=Open-source / open-weight models2026.06 | 49.8 | 35.4 | |
| Llama 4 MaverickSource type=Open-source / open-weight models2026.06 | 46.1 | 34.8 | |
| Qwen3-4BSource type=Open-source / open-weight models2026.06 | 45.2 | 31.8 |