AgentWorldBench Web
AgentWorldBench Web domain - open-ended rubric evaluation for language world models simulating web browser environments. Normalized to 0-100.
Category
General Agents
Max Score
100.0
Score Type
percent
Active
Yes
Model Scores
Scores for AgentWorldBench Web
| Model | Score | Date | Verified |
|---|---|---|---|
| Qwen-AgentWorld-35B-A3B |
32.41%
|
24.06.2026 | Verified |
| Qwen-AgentWorld-397B-A17B |
51.32%
|
24.06.2026 | Unverified |
| Qwen3.5-35B-A3B |
47.10
|
24.06.2026 | Unverified |
| Qwen3.5-397B-A17B |
19.18%
|
24.06.2026 | Unverified |
| Qwen3.6-Plus |
48.68%
|
24.06.2026 | Unverified |
| GPT-5.4 |
62.17%
|
24.06.2026 | Unverified |
| Claude Opus 4.8 |
100.00%
|
24.06.2026 | Unverified |
| Claude Opus 4.6 |
57.14%
|
24.06.2026 | Unverified |
| Gemini 3.1 Pro |
75.79%
|
24.06.2026 | Unverified |
| Claude Sonnet 4.6 |
48.68%
|
24.06.2026 | Unverified |
| DeepSeek-V4-Pro |
42.59%
|
24.06.2026 | Unverified |
| GLM-5.1 |
58.20%
|
24.06.2026 | Unverified |
| Kimi K2.6 |
41.01%
|
24.06.2026 | Unverified |
| MiniMax-M2.7 |
45.24%
|
24.06.2026 | Unverified |