V2X-QA - Cooperative Average: leaderboard

V2X-QA (2026) asks multimodal models four-option driving questions about real vehicle-mounted and roadside camera views of the same scenes, answered with one option letter on the expert-verified test split. This board is the mean accuracy of the four cooperative tasks, which give the model both views. Accuracy in percent; higher is better.

Metric: Accuracy (%) on the cooperative view (vehicle and roadside images together), mean of its four tasks; four-option multiple-choice questions on the expert-verified test split, answered as a single option letter under the paper's unified prompt; higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Intern-S143.5#624
2Qwen 2.5 VL 72B Instruct42#841
3GPT-5.236.7#185
4Gemini 3 Flash (Preview)36.7#152
5Intern-S1-mini35.6#1178
6Qwen 3.5 Plus34#242
7Gemini 2.5 Pro30.9#276
8Qwen3-VL-Flash30.4#728
9Qwen 3 VL 30B A3B Instruct27.5#895
10GPT-5 Mini (2025-08-07)23.7#345

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=v2x-qa-cooperative-average · How It Works · Data refreshed daily, snapshot 2026-11-21.