AJ-Bench (LLM-as-a-Judge) - Wide Search: leaderboard
Metric: F1 (%) on Wide search (WideSearch-derived item checks; 9 tasks, 27 trajectories), averaged over three runs, with the judge reading the trajectory only (no tools or environment access), of AJ-Bench, judging whether agent trajectories succeeded against binary human-verified labels (155 tasks, 516 trajectories), default model configurations; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 72.7 |
| 2 | Grok 4 | 69.18 |
| 3 | GLM-4.6 | 66.61 |
| 4 | Gemini 2.5 Pro | 66.35 |
| 5 | GPT-5 | 66.33 |
| 6 | LongCat-Flash-Chat | 64.44 |
| 7 | Claude Opus 4.5 | 64.26 |
| 8 | DeepSeek V3.2 | 63.65 |
| 9 | Kimi K2 0905 | 63.52 |
| 10 | Qwen 3 235B A22B | 62.69 |
| 11 | Claude Sonnet 4.5 | 61.02 |
| 12 | GPT-5 Mini (Low) | 60.84 |
| 13 | GPT-5.1 | 58.02 |
Interactive version: theaggregate.ai/benchmark?slug=aj-bench-llm-as-a-judge-wide-search · How It Works · Data refreshed daily, snapshot 2026-10-07.