LLM Stats (Video-MME): leaderboard
900 videos spanning 6 domains with 2,700 QA pairs testing short/medium/long video understanding with optional subtitles.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Seed 2.1 Pro | 89.2 | #67 |
| 2 | Seed 2.1 Turbo | 89 | #86 |
| 3 | Qwen 3.7 Plus | 88 | #92 |
| 4 | MiMo-V2.5 | 87.7 | #196 |
| 5 | Kimi K2.5 | 87.4 | #144 |
| 6 | MiniMax-M3 | 85.4 | #136 |
| 7 | Gemini 2.5 Pro | 84.8 | #152 |
| 8 | Qwen 3.6 Plus | 84.2 | #120 |
| 9 | Gemini 1.5 Pro | 78.6 | #445 |
| 10 | Gemini 1.5 Flash | 76.1 | #618 |
| 11 | Qwen 3 VL 30B A3B Instruct | 74.5 | #384 |
| 12 | Qwen 3 VL 30B A3B (Thinking) | 73.3 | #351 (Qwen 3 VL 30B A3B) |
| 13 | Qwen 3 VL 8B (Thinking) | 71.8 | |
| 14 | Qwen 3 VL 8B Instruct | 71.4 | #420 |
| 15 | Gemini 1.5 Flash-8B | 66.2 | #819 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-video-mme · How It Works · Data refreshed daily, snapshot 2026-10-20.