See-or-Guess VQA - Subquestion-Guided: leaderboard
Metric: Subquestion-Guided variant (built from single-pass perceptual sub-questions): accuracy (%) with the image over 540 images from 21 public benchmarks in six reasoning categories; Qwen3-14B judges answer equivalence (yes/no), lenient extraction of answers after a missing </think> tag. Source: arxiv.org. Saturation forecast: Around July 2027. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 9B | 56.8 |
| 2 | Qwen 3.5 397B A17B | 54.6 |
| 3 | Gemini 2.5 Pro | 53.3 |
| 4 | GPT-5 Mini | 49.4 |
| 5 | Claude Sonnet 4.6 | 47.8 |
| 6 | Gemini 3.1 Flash Lite | 41.8 |
| 7 | InternVL3.5-8B | 28.9 |
| 8 | Phi-4 Multimodal Instruct | 22.6 |
Interactive version: theaggregate.ai/benchmark?slug=see-or-guess-vqa-subquestion-guided · How It Works · Data refreshed daily, snapshot 2026-09-29.