VerdictBench: leaderboard
E-commerce platforms have begun recruiting crowdsourced jurors to adjudicate massive volumes of transaction disputes. Unlike formal legal judgment, E-commerce dispute verdicts require grounding pivotal clues from redundant, multi-round, multimodal evidence and making decisions under flexible platform-specific conventions. These characteristics render existing methods insufficient for this scenario
Metric: Accuracy (self-reported). Source: benchmarklist.com. Saturation forecast: Around April 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 63.54 | #80 |
| 2 | GPT-5.2 Instant | 63.44 | #210 |
| 3 | DeepSeek V3 | 60.8 | #321 |
| 4 | Qwen Plus | 60.55 | #385 |
| 5 | Claude Opus 4.5 | 59.1 | #83 |
| 6 | Seed-1.6 | 56.26 | #265 |
| 7 | Grok 4 | 54.36 | #181 |
| 8 | Gemini 2.5 Flash Lite | 52.92 | #422 |
| 9 | Llama 2 13B | 50.42 | #1436 |
| 10 | GLM-4 | 50.17 | #648 |
| 11 | Qwen 3 VL 235B A22B | 48.43 | #238 |
| 12 | GPT-5.2 | 48.33 | #110 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=verdictbench · How It Works · Data refreshed daily, snapshot 2026-10-20.