Skip to content
Newsroom
Models 3d ago by Rajat Jain

9 of Top 10 Text-to-Video Models Are Chinese

Bloomberg says nine of the top ten text-to-video models on Artificial Analysis are now Chinese — outside Google, ByteDance's Seedance 2.5 and MiniMax's H3 lead.

9 of Top 10 Text-to-Video Models Are Chinese

Nine of the top ten text-to-video systems on Artificial Analysis’ leaderboard are Chinese — outside of Google, the video-generation frontier has effectively moved to China. A Bloomberg Opinion column published August 9 frames video as the clearest evidence of China’s AI prowess, with the stakes extending well beyond Hollywood into world-model buildouts.

The releases driving the gap: ByteDance’s Seedance 2.5 and MiniMax’s H3, both shipping in recent weeks as the capability layer of a video stack that also feeds robotics simulation and autonomous-world modeling.

Key facts

  • The stat: 9 of top 10 text-to-video models on Artificial Analysis’ leaderboard are Chinese (Google being the non-Chinese entry).
  • The leaders: ByteDance’s Seedance 2.5, MiniMax’s H3 — plus a deep Chinese bench behind them.
  • Why video matters beyond content: text-to-video is the nearest public proxy for world models — the simulation layer for robotics, autonomous driving, and agent training environments.
  • US position: Google holds one top-10 slot (Veo lineage); OpenAI’s Sora-era position has not held the field’s top tier through 2026 per this analysis.
  • The framing: Bloomberg’s column argues video is where the US-China gap is widest — and the most strategically under-appreciated.

Why it matters

  • World models are the next hundred-billion-dollar layer. Whoever controls video generation controls the simulation substrate for physical AI — this is not an entertainment story.
  • The Hollywood angle is the visible tip. Licensing, studio adoption, and creator workflows are where the gap shows in headlines; robotics simulation is where it compounds.
  • US labs’ position is defensive, not absent. Google’s one top-10 slot and Veo’s enterprise momentum are the containment story; nobody is catching the Chinese bench this quarter.

What to watch

  • Independence of the leaderboard’s next update — whether US entries regain ground on quality or price.
  • The enterprise simulation pipeline — Chinese video models feeding robotics sim (the Shenzhen deployment story connects directly to this).
  • Model-level response from Google (Veo) and any OpenAI video push — the window for a US counter-move.

Official source

Updated August 10, 2026.

#Models #Video #China #ByteDance