FGTR-Bench - Roadside Micro-Objects: leaderboard
The roadside micro-object subset of FGTR-Bench (2026): accuracy in percent on 800 4-option multiple-choice questions about small roadside hazards such as cones and debris, zero-shot. Higher is better.
Metric: Accuracy (%) on 800 roadside tiny-object 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 February 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 VL 7B Instruct | 34 | #1833 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=fgtr-bench-roadside-micro-objects · How It Works · Data refreshed daily, snapshot 2026-11-13.