EVID-Bench - Magnitude Manipulation: leaderboard

Metric: Point-level accuracy (%) on the 23 magnitude manipulation videos (multi-source editing: footage spliced to exaggerate scale), using the retrieval-augmented verification pipeline of the paper (chain-of-thought analysis, up to six YouTube search-verify-reflect rounds, 64 frames at 480p, temperature 0); each video has 3 to 5 annotated misinformation points, matched by a majority of three LLM judges; mean of three runs; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.

Top models

#ModelScore
1GPT-5.598.91
2Claude Opus 4.696.74
3GPT-5.495.65
4Claude Sonnet 4.694.57
5Qwen 3.5 Plus88
6GPT-5.4 Mini85.87
7Gemini 3.1 Pro (Preview)82.6
8Gemini 3 Flash80.4
9Qwen 3 VL 235B A22B Instruct73.9

Interactive version: theaggregate.ai/benchmark?slug=evid-bench-magnitude-manipulation · How It Works · Data refreshed daily, snapshot 2026-09-29.