ByteDance plans to launch Dreamina Seedance 2.5 in July, extending AI video generation from 15-second clips to 3-minute sequences.

The model will debut across Dreamina, CapCut, and partner platforms that integrate ByteDance's video stack. Public product pages already reference the upgrade, though the July 9 launch date remains unconfirmed.

Seedance 2.5 introduces a tiered generation system: 30-second standard scenes, 90-second drafts, and 180-second final outputs. The workflow represents ByteDance's biggest duration leap since launching the original model.

Technical challenges ahead

The key test will be maintaining character consistency, motion logic, and prompt adherence across extended sequences. Current Seedance 2.0 handles 4-15 second clips with multimodal references and audio-video sync.

For creators, advertisers, and AI filmmakers, the upgrade transforms Seedance from a short-clip generator into a tool for commercial storytelling. The longer format directly targets workflows in social media, advertising, and animation production.

ByteDance has been integrating AI video across its creator ecosystem through Dreamina, CapCut, and TikTok production tools. The company distributes Seedance through BytePlus APIs alongside its consumer applications.

The timing puts ByteDance ahead of competitors after OpenAI discontinued Sora's web and app products in April. Google's Veo and Gemini Omni remain the primary competitive pressure in extended AI video generation.

Seedance 2.5 will roll out through Dreamina's existing interface, CapCut's creator tools, and third-party applications with ByteDance model access.