V2X-QA - RSU Traffic Agents: 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 roadside-unit task on traffic participants over the wide field of view. Accuracy in percent; higher is better.

Metric: Accuracy (%) on the infrastructure-side RSU Traffic Agents task (IS2); 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: Not forecast. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Intern-S161.5#624
2Qwen 2.5 VL 72B Instruct58.1#841
3Qwen 3.5 Plus56.6#242
4Intern-S1-mini55.1#1178
5GPT-5 Mini (2025-08-07)52.1#345
6Qwen3-VL-Flash50#728
7Qwen 3 VL 30B A3B Instruct45.7#895
8Gemini 3 Flash (Preview)44.9#152
9Gemini 2.5 Pro43#276
10GPT-5.242.6#185

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-rsu-traffic-agents · How It Works · Data refreshed daily, snapshot 2026-11-21.