VAREX-Bench — leaderboard
Multi-modal structured extraction benchmark over 1,777 government forms with per-document JSON schemas and image/text/spatial modalities.
Metric: Exact Match (self-reported). Source: benchmarklist.com. Status: saturation imminent. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 98 |
| 2 | Gemini 2.5 Flash | 97.3 |
| 3 | Llama 4 Maverick | 95.6 |
| 4 | GPT-4o | 94.8 |
| 5 | Llama 4 Scout | 94.3 |
Interactive version: theaggregate.ai/benchmark?slug=varex-bench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.