CiteVQA - Single-Doc: leaderboard

CiteVQA (2026) asks multimodal models to answer questions over long PDFs, 1,897 questions on 711 documents in seven domains, and to cite the exact page elements that support each answer as bounding boxes. This board covers the questions on a single document, counting a question as correct only when a judge scores the answer at least 4 of 5 and the cited evidence is either judged relevant (at least 4 of 5) or recovers at least 60 percent of the key evidence; higher is better.

Metric: Strict attributed accuracy (%): share of questions with answer score at least 4 and either relevance at least 4 or evidence recall at least 0.6, on the single-document questions (52.0 percent of the set), 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 February 2027. 20 models tracked. Updated 14 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.1 Pro (Preview)High76#102
2Gemini 3 Flash (Preview)High69.3#156
3GPT-5.4xHigh61.7#138
4Seed 2.0 Pro51.9#226
5Gemini 2.5 ProThinking, 32K49.4#280
6GPT-5.2xHigh32.6#190
7Qwen 3 VL 235B A22B Instruct25#541
8Kimi K2.521.3#272
9Qwen 3.6 Plus20.2#245
10Qwen 3 VL 32B Instruct19.3#607
11Qwen 3.5 397B A17B17.7#270
12Qwen 3.5 27B17.1#380
13Gemma 4 31B16.4#390
14Qwen 3.5 122B A10B16#350
15GLM-5V Turbo14.1#365

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=citevqa-single-doc · How It Works · Data refreshed daily, snapshot 2026-11-22.