AgentWorldBench OS
AgentWorldBench OS domain - open-ended rubric evaluation for language world models simulating operating-system environments. Normalized to 0-100.
Category
General Agents
Max Score
100.0
Score Type
percent
Active
Yes
Model Scores
Scores for AgentWorldBench OS
| Model | Score | Date | Verified |
|---|---|---|---|
| Qwen-AgentWorld-35B-A3B |
69.27%
|
24.06.2026 | Verified |
| Qwen-AgentWorld-397B-A17B |
83.42%
|
24.06.2026 | Unverified |
| Qwen3.5-35B-A3B |
56.27
|
24.06.2026 | Unverified |
| Qwen3.5-397B-A17B |
32.88%
|
24.06.2026 | Unverified |
| Qwen3.6-Plus |
29.15%
|
24.06.2026 | Unverified |
| GPT-5.4 |
88.37%
|
24.06.2026 | Unverified |
| Claude Opus 4.8 |
74.30%
|
24.06.2026 | Unverified |
| Claude Opus 4.6 |
100.00%
|
24.06.2026 | Unverified |
| Gemini 3.1 Pro |
76.45%
|
24.06.2026 | Unverified |
| Claude Sonnet 4.6 |
49.53%
|
24.06.2026 | Unverified |
| DeepSeek-V4-Pro |
53.34%
|
24.06.2026 | Unverified |
| GLM-5.1 |
20.53%
|
24.06.2026 | Unverified |
| Kimi K2.6 |
32.52%
|
24.06.2026 | Unverified |
| MiniMax-M2.7 |
10.48%
|
24.06.2026 | Unverified |