VerdictBench — leaderboard
Metric: Accuracy (self-reported). Source: benchmarklist.com. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3 | 60.8 |
| 2 | Claude Opus 4.5 | 59.1 |
| 3 | Grok 4 | 54.36 |
| 4 | Gemini 2.5 Flash Lite | 52.92 |
| 5 | Llama 2 13B | 50.42 |
| 6 | Qwen 3 VL 235B A22B | 48.43 |
| 7 | GPT-5.2 | 48.33 |
Interactive version: theaggregate.ai/benchmark?slug=verdictbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.