FireWorldBench - Real-World-Aligned (Rendered Images): leaderboard

The real-world-aligned subset of FireWorldBench (Chen et al., 2026): 26 recorded indoor fire experiments reconstructed with the Fire Dynamics Simulator, with questions across the five physical-capability dimensions. Observations are given as physics-rendered field images and 3D scene renderings; the score is completion accuracy in percent averaged over the five dimensions and over matched multiple-choice and open-report questions and scored deterministically; higher is better.

Metric: Completion accuracy (%; equal-weight mean of P1-P5 on the 714-item real-world-aligned subset, 26 event groups reconstructed from recorded indoor fire experiments; image-observation track (physics-rendered field visualizations and 3D scene renderings); each cell is the mean of the matched choice-format score (Jaccard overlap of selected and gold option sets) and open-report score (exact agreement of required structured slots); deterministic evaluator, no LLM judge; failed or unparseable responses score 0). Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.

Top models

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=fireworldbench-real-world-aligned-rendered-images · How It Works · Data refreshed daily, snapshot 2026-11-12.