FGTR-Bench - Daytime Signal: leaderboard

Metric: Accuracy (%) on 750 daytime traffic-signal questions, 4-option multiple-choice questions on the blind test split, zero-shot (models without FGTR-Bench training), greedy letter decoding; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 5 models tracked.

Top models

#ModelScore
1Qwen 2.5 VL 7B Instruct97.9

Interactive version: theaggregate.ai/benchmark?slug=fgtr-bench-daytime-signal · How It Works · Data refreshed daily, snapshot 2026-09-29.