Causal-Plan-Bench - Plan Diagnosis and Repair: leaderboard

Metric: Open-ended bad plan diagnosis and repair (find the wrong step and repair the plan) task, rubric score (0-100): a blind GPT-5.6-sol judge scores each open-ended answer against a human-verified reference from 0 to 1, times 100; 100 held-out expert-reviewed items per task drawn from egocentric and robot video (EPIC-KITCHENS, Ego4D, HoloAssist and others), the same instruction, visual evidence and answer format for every model, temperature 0.2, mean of three regenerated runs; higher is better. Source: arxiv.org. Saturation forecast: Around 2033. 14 models tracked.

Top models

#ModelScore
1GPT-639.8
2GPT-5.6 Sol38.3
3Gemini 3.1 Pro (Preview)38.2
4Kimi K2.536.9
5Seed 2.1 Pro36.9
6Qwen 3 VL 8B Instruct30.5
7GPT-4o25.4
8InternVL3.5-8B25.4

Interactive version: theaggregate.ai/benchmark?slug=causal-plan-bench-plan-diagnosis-and-repair · How It Works · Data refreshed daily, snapshot 2026-09-29.