VeriTaS: The First Dynamic Benchmark for Multimodal Automated Fact-Checking: leaderboard
VeriTaS reports source-table results for multimodal automated fact-checking systems on real-world claims collected from fact-checking organizations. The benchmark covers textual and audiovisual content, and the pinned metric is MSE, measuring the model’s scoring error against the benchmark’s standardized verdict-and-justification scheme. It is a dynamic benchmark updated quarterly. Source: BenchmarkList’s attributed result table.
Metric: MSE. Source: benchmarklist.com. Saturation forecast: Around June 2027. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 0.18 | #67 |
| 2 | Gemini 3 Flash | 0.28 | #103 |
| 3 | Qwen 3.5 397B A17B | 0.32 | #154 |
| 4 | Gemma 4 31B | 0.36 | #212 |
| 5 | Gemini 3.1 Pro (Preview) | 0.39 | #64 |
| 6 | GPT-5 | 0.83 | #105 |
| 7 | Llama 4 Maverick | 0.86 | #486 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=veritas-the-first-dynamic-benchmark-for-multimodal-automated-fact-checking · How It Works · Data refreshed daily, snapshot 2026-11-05.