CiteVQA - Answer Correctness: leaderboard
Metric: Answer correctness against the reference answer, judged on a 0-5 scale and multiplied by 20 (0-100), over all 1,897 questions, the model answers from the document pages and cites element-level bounding boxes, unified prompt, temperature 1.0, at most 4,096 output tokens, Qwen3-VL-235B-A22B judge; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 (xHigh) | 87.1 |
| 2 | Gemini 3.1 Pro (Preview) (High) | 86.1 |
| 3 | Qwen 3.6 Plus | 85.9 |
| 4 | Gemini 3 Flash (Preview) (High) | 84.5 |
| 5 | Seed 2.0 Pro | 81.3 |
| 6 | Qwen 3.5 397B A17B | 76.5 |
| 7 | Qwen 3.5 35B A3B | 76.4 |
| 8 | Qwen 3.5 27B | 75.6 |
| 9 | Kimi K2.5 | 74.3 |
| 10 | Qwen 3.5 122B A10B | 73.6 |
| 11 | Qwen 3 VL 32B Instruct | 72.3 |
| 12 | Qwen 3 VL 235B A22B Instruct | 72.3 |
| 13 | GPT-5.2 (xHigh) | 71.5 |
| 14 | Gemma 4 31B | 69.8 |
| 15 | Qwen 3.5 9B | 65 |
Interactive version: theaggregate.ai/benchmark?slug=citevqa-answer-correctness · How It Works · Data refreshed daily, snapshot 2026-10-07.