Real-World Agent on Claw-Eval
80.6Average ScoreGPT-5.4
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GPT-5.42026.03 | 80.6 | 66.3 | |
| Claude Opus 4.62026.03 | 79.3 | 66.3 | |
| SkillAdaptorBackbone=GPT-5.22026.05 | 77.6 | 78.9 | |
| OpenSpaceBackbone=GPT-5.22026.05 | 77.1 | 77.4 | |
| Base ModelBackbone=GPT-5.22026.05 | 76 | 77.4 | |
| SkillAdaptorBackbone=Kimi-K2.52026.05 | 75.8 | 77.4 | |
| SkillAdaptorBackbone=GLM-52026.05 | 75.3 | 75.9 | |
| Gemini 3.1 Pro2026.03 | 74.2 | 50 | |
| OpenSpaceBackbone=Kimi-K2.52026.05 | 74 | 75.9 | |
| OpenSpaceBackbone=GLM-52026.05 | 73.6 | 74.4 | |
| KAT-Coder-V22026.03 | 73.4 | 55.6 | |
| Base ModelBackbone=Kimi-K2.52026.05 | 73.2 | 74.4 | |
| GLM-52026.03 | 73 | 57.7 | |
| Base ModelBackbone=GLM-52026.05 | 72.8 | 72.9 | |
| MiniMax M2.72026.03 | 70.7 | 51.9 | |
| Qwen3.5Parameter Count=9B2026.05 | 65.4 | — | |
| Gemma4Parameter Count=26BA4B2026.05 | 60.6 | — | |
| SenseNova-U1Parameter Count=8B, Configuration Variant=Think2026.05 | 58.9 | — | |
| SenseNova-U1Parameter Count=30BA3B, Configuration Variant=Think2026.05 | 58.5 | — | |
| Qwen3.5Parameter Count=35BA3B2026.05 | 36.5 | — | |
| Qwen3VLParameter Count=30BA3B, Configuration Variant=Think2026.05 | 22.1 | — | |
| Qwen3VLParameter Count=8B, Configuration Variant=Think2026.05 | 21.7 | — |