EVID-Bench: leaderboard
EVID-Bench (2026) gives a multimodal model a manipulated video inside a fixed search-and-compare pipeline; it must find related source videos on the web and state what information is false. Scored by point-level accuracy in percent over 222 videos with 3 to 5 annotated misinformation points each, matched by a majority of three LLM judges; higher is better.
Metric: Point-level accuracy (%) over all 222 manipulated videos: share of ground-truth misinformation points the model identifies, 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: Around May 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 61.43 | #49 |
| 2 | Claude Opus 4.6 | 58.34 | #102 |
| 3 | Gemini 3.1 Pro (Preview) | 55.9 | #94 |
| 4 | Claude Sonnet 4.6 | 55.13 | #161 |
| 5 | GPT-5.4 | 53.77 | #123 |
| 6 | Gemini 3 Flash | 53.6 | #164 |
| 7 | Qwen 3.5 Plus | 53.4 | #232 |
| 8 | Qwen 3 VL 235B A22B Instruct | 48.8 | #522 |
| 9 | GPT-5.4 Mini | 46.97 | #389 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=evid-bench · How It Works · Data refreshed daily, snapshot 2026-11-13.