BBC News Video Captioning - Text Similarity: leaderboard
Metric: Text similarity (cosine, -1 to 1, times 100) between all-MiniLM-L12-v2 sentence embeddings of the generated caption and the reference, averaged over clips, the authors' BBC News corpus of 9,838 news video clips (at most 300 seconds) with their editorial descriptions as references; each open-source video LLM writes a free-form description of the clip from sampled frames; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 12B (IT) | 43.87 | #655 |
| 2 | Qwen 2.5 VL 7B Instruct | 36.07 | #643 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=bbc-news-video-captioning-text-similarity · How It Works · Data refreshed daily, snapshot 2026-10-11.