ChronoPhyBench - Next-State Selection (3 Options): leaderboard

Metric: Accuracy (%) on ChronoPhyBench next-state frame selection with 3 candidates: given a real physical-interaction video context (8 uniformly sampled frames) and its caption, choose the frame that comes next; closed models at temperature 0.1, open models at their recommended sampling; chance 33.3; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 22 models tracked.

Top models

#ModelScore
1GPT-5.572.5
2Gemini 3 Pro (Preview)68.5
3Doubao-Seed-2.0-Pro-26021568
4Doubao-Seed-1-8-25122855
5GPT-5.452.5
6GLM-4.1V-9B (Thinking)52.5
7GPT-5.345.5
8Doubao-Seed-1.6-thinking-25071541.5
9InternVL3-38B41
10InternVL3-14B40.5
11Qwen 3 VL 8B (Thinking)40
12Qwen 3 VL 8B Instruct38.5
13InternVL3-8B37.5
14Qwen 3 VL 32B Instruct37
15InternVL3.5-8B30

Interactive version: theaggregate.ai/benchmark?slug=chronophybench-next-state-selection-3-options · How It Works · Data refreshed daily, snapshot 2026-09-29.