New AA-Video-T2V v2.0 Benchmark Ranks Wan 3.0 #1, MiniMax H3 at 1/7 the Price of Seedance
ArtificialAnlys · x · 2026-09-30
Artificial Analysis launched AA-Video-T2V v2.0 and its Silent variant, new text-to-video benchmarks judging every model at 1080p on a refreshed prompt set, built from 68,000+ human preference votes over 1,000 prompts across 10 use cases, 10 capabilities, and multiple visual styles.
Key leaderboard findings:
- Wan 3.0 ranks #1 (Elo 1157, $12/min), leading 10 of 20 category boards including Cartoon/Anime styles and Animation & Gaming
- Dreamina Seedance 2.5 is #2 (1143, $34.12/min — the priciest), #1 on Human Anatomy and Dialogue & Lip Sync
- MiniMax H3 ranks #3 (1138, $4.80/min, 1/7 of Seedance's price, statistically tied), #1 on Text Rendering
- FLUX 3 is #4, strong on Text Rendering (#3) and Lip Sync (#2)
- Gemini Omni Flash 1.1 is #5, the specialist for UI/UX & Motion Design, Flat Design, and Audio Sync (77/62 Elo above its overall score)
The methodology emphasizes balanced coverage of real-world use cases and capabilities, reflecting versatility rather than single strengths. The authors note AI video is moving into microdramas and theaters while low barriers flood feeds with low-quality content, so the quality bar keeps rising.
More from Multimodal
- Testing H3 orbit finetunes: 360 Orbit LoRA, ORB360 CardSpin and Meridian, no clear winner — linoy_tsaban · 2026-09-30
- 1,337 Copy-Paste Chart Style Prompts for Codex and Claude Code, Built With Opus — mhmazur · 2026-09-30
- Dev Visualizes 'Bangmaid' Idea with NoSpoon Microdrama Agent, Delighting Peers — Kyrannio · 2026-09-30
- Runway shows how to turn one product photo into a full social ad campaign — tlakomy · 2026-09-30
- Pencil0 enters private beta: an AI filmmaking agent that turns one logline into a full film — yihui_indie · 2026-09-30
- SoL-Refiner turns low-res AI video into 2K/4K in one denoising step, 8.91x faster — linoy_tsaban · 2026-09-30