Seedance 2.5

Seedance 2.5 is ByteDance Seed’s next-generation audio-video joint-generation model. The official product page describes single-pass video generation of up to 30 seconds, with two possible extensions, smoother motion, more consistent visuals, and native audio-video generation. It supports multimodal reference inputs and more precise editing, including reference video understanding, white-model control, green-screen editing, camera movement, and performance blocking. The model is a foundation capability rather than a standalone consumer subscription; the official announcement says it is rolling out through Jimeng AI, Doubao Pro, and other platforms, with API access now available through BytePlus ModelArk. The official provider documents token billing and activation requirements for Dreamina Seedance 2.5.

✨ Key Features

• 30-second generation — Create longer narrative clips in one generation.
• Multi-round extension — Extend generated content for longer stories.
• Joint audio-video generation — Generate synchronized visual and audio output.
• Multimodal references — Use image, video, and audio references to guide output.
• Reference video understanding — Capture framing, motion, and cinematic intent.
• Timestamp-level editing — Target changes to specific moments in a clip.
• Green-screen editing — Support compositing-oriented video workflows.
• White-model control — Use structure and blocking references for complex scenes.
• Camera control — Guide camera perspective and movement.
• Performance blocking — Coordinate subjects, actions, and shot transitions.
• Improved continuity — Maintain more consistent scenes and motion across shots.

💳 Pricing

• API access — The current official model page exposes Get API and Playground entry points. BytePlus ModelArk now offers Dreamina Seedance 2.5, model ID dreamina-seedance-2-5-260628. It is token-billed rather than a standalone consumer subscription.
• 480p/720p — USD $10.70/million tokens without an input video, or $6.40/million with an input video.
• 1080p — USD $11.70/million tokens without an input video, or $7.00/million with an input video. Cost also varies with resolution, input/output duration and actual token consumption; a lower unit rate does not guarantee a cheaper video.
• Generation terms — Successful generations are billed; failed or moderation-rejected generations are not. Draft and final-generation steps are billed independently.
• Activation — The current guide requires verified Credits of at least $30, or a qualifying AI Savings Plan/resource pack of at least $30 with a remaining balance/quota. Plans and resource packs have expiry and cancellation terms.
• Trial boundary — Generic ModelArk free-inference materials do not confirm a Seedance 2.5-specific free allowance. Do not infer a free trial from the Playground link; other distribution platforms have their own terms.

⚖️ Alternatives Compared

🎯 Use Cases

• Short-film production: Create narrative scenes with synchronized audio.
• Commercial concepts: Prototype advertising shots and transitions.
• Reference-driven video: Guide output with images, clips, and audio.
• Storyboard development: Turn shot directions into a visual sequence.
• Character continuity: Maintain subjects across complex scenes.
• Green-screen composites: Generate footage for compositing workflows.
• Camera blocking: Explore controlled camera motion and framing.
• API experiments: Test audio-video generation through the live provider API.