SteelBench - Compositional Reasoning Gap: leaderboard

Metric: Compositional reasoning gap (0-1; share of worker instances with a correctly recognised action whose safety-violation judgment is still wrong, lower is better; 1,345 densely annotated clips of real steel-plant CCTV footage (8 frames at 1080p per clip) from six industrial domains, zero-shot structured prompt with the 25-class action taxonomy, site context, PPE definitions and safety rules; provider-default temperature; failed parses count as wrong). Source: arxiv.org. Saturation forecast: Around 2031. 9 models tracked.

Top models

#ModelScore
1Llama 4 Maverick0.38
2GPT-4o0.39
3GPT-5.40.48
4Gemini 2.5 Pro0.5
5Gemini 2.5 Flash0.5
6Nemotron Nano 12B v2 VL0.52
7Gemma 4 31B (IT)0.55
8Claude Opus 4.70.57
9Qwen 3.5 122B A10B0.58

Interactive version: theaggregate.ai/benchmark?slug=steelbench-compositional-reasoning-gap · How It Works · Data refreshed daily, snapshot 2026-09-29.