ParsiEval (279 Questions): leaderboard
Metric: Accuracy (%, 279 questions). Source: github.com. 1 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | qwen3-4b-thinking-2507 | 51.97 |
Interactive version: theaggregate.ai/benchmark?slug=parsieval-279-questions · How It Works · Data refreshed daily, snapshot 2026-09-20.