FireWorldBench - Temporal Evolution Forecasting (Rendered Images): leaderboard

Metric: Completion accuracy (%; P1, temporal evolution forecasting items of the controlled-simulation subset; 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 December 2026. 10 models tracked.

Top models

#ModelScore
1GPT-5.6 Luna60.74
2Qwen 3 VL 32B Instruct58.28
3Claude Haiku 4.556.04
4Gemini 2.5 Flash52.84
5InternVL3-78B45.96
6Qwen 3.5 Plus (2026-02-15)44.92
7InternVL3-38B41.94
8Qwen 3 VL 235B A22B Instruct38.02
9Qwen 3 VL 8B Instruct31.92
10InternVL3-8B26.86

Interactive version: theaggregate.ai/benchmark?slug=fireworldbench-temporal-evolution-forecasting-rendered-images · How It Works · Data refreshed daily, snapshot 2026-09-26.