FireWorldBench - Real-World-Aligned (Rendered Images): leaderboard
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 September 2027. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Luna | 51.51 |
| 2 | Gemini 2.5 Flash | 47.65 |
| 3 | Qwen 3 VL 32B Instruct | 47.38 |
| 4 | Claude Haiku 4.5 | 45.95 |
| 5 | Qwen 3.5 Plus (2026-02-15) | 43.47 |
| 6 | Qwen 3 VL 235B A22B Instruct | 41.47 |
| 7 | InternVL3-78B | 38.47 |
| 8 | Qwen 3 VL 8B Instruct | 36.68 |
| 9 | InternVL3-38B | 36.62 |
| 10 | InternVL3-8B | 26.93 |
Interactive version: theaggregate.ai/benchmark?slug=fireworldbench-real-world-aligned-rendered-images · How It Works · Data refreshed daily, snapshot 2026-09-26.