ERQA-Plus: leaderboard

Metric: Multiple-choice accuracy (%) over all 1,766 questions in 12 reasoning subtypes; multiple-choice questions over single and multi-image embodied scenes, generated by GPT-4o and filtered by a Gemini 3 Pro judge agent over 1,766 questions; the model outputs only the option letter, read by regular expression; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 7 models tracked.

Top models

#ModelScore
1Qwen 3 VL 32B (Thinking)83.4

Interactive version: theaggregate.ai/benchmark?slug=erqa-plus · How It Works · Data refreshed daily, snapshot 2026-09-29.