ByteDance's Seedance 2.0 has taken the top spot on the Artificial Analysis video leaderboards, beating Google's Veo 3, OpenAI's Sora, and Kling across both text-to-video and image-to-video categories. The Elo scores tell the story: 1269 for text-to-video (without audio), 1350 for image-to-video. Those rankings come from blind user voting, not ByteDance's own benchmarks.
Access is another matter. The model launched on February 12, 2026, and immediately went viral with users generating clips of Hollywood characters in absurd scenarios. Disney, Paramount, and the MPA fired off cease-and-desist letters within days. ByteDance responded by suspending the face-to-voice feature, blocking real human face uploads in image-to-video, and on March 15, officially suspending the overseas API through BytePlus.
So who can actually use it? In mainland China, Jimeng and Xiaoyunque still serve the model. Internationally, Dreamina (ByteDance's CapCut-linked platform) is the intended gateway, but Seedance 2.0 access there remains limited. Some users on Twitter report getting through with Southeast Asian IPs. The global CapCut app doesn't have 2.0 yet. Third-party API providers that briefly offered access have mostly pulled it.
The face-blocking issue is real and frustrating creators. ByteDance's content filter rejects uploaded photos containing identifiable human faces before the prompt even processes, a direct response to the deepfake controversy. Workarounds circulating online involve AI-generated portrait references instead of real photos.
As for the question of whether appearing on Artificial Analysis means API access exists: not exactly. The Arena leaderboard is populated through blind comparisons, and models can be evaluated there without having a public API. The official overseas API has no confirmed relaunch date.
Bottom Line
Seedance 2.0 leads both Artificial Analysis video leaderboards by Elo score, but ByteDance suspended its overseas API on March 15 due to copyright disputes, leaving international access fragmented.
Quick Facts
- Elo 1269 for text-to-video, 1350 for image-to-video on Artificial Analysis (user-voted)
- Launched February 12, 2026 by ByteDance
- Overseas API officially suspended March 15, 2026
- Real human face uploads blocked in image-to-video mode
- Dreamina international access limited; no confirmed public rollout date
