CAGE - Intervention (Chain-Text Probe) - Chain Score: leaderboard

Metric: Judge score (0-10) for the causal chain in arrow notation, mean of GPT-4o and Claude 3.5 Sonnet ratings against GPT-4o-generated reference answers and chains; Chain-Text Probe, level-2 intervention questions (what happens if something in the image changes), where the chain needs at least one causal link; 500 held-out COCO images, three questions each; blank responses score 0. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 8 models tracked.

Top models

#ModelScore
1LLaVA-NeXT-13B7.83
2InternVL2-76B3.9
3Qwen-VL-Chat3.77
4MiniGPT-4-Vicuna-13B2.09
5LLaVA-CoT0.97
6LLaVA-RLHF-13B0.81
7CogVLM2-19B0.54
8mPLUG-Owl20.42

Interactive version: theaggregate.ai/benchmark?slug=cage-intervention-chain-text-probe-chain-score · How It Works · Data refreshed daily, snapshot 2026-09-29.