Drafted with AI tools and edited by a human. The figures and conclusions were checked by a GPUniq editor.
TL;DR
The Dreamina Seedance 2.5 Flash API costs $0.152 per generated second for 720p output. It runs nearly two times faster than the flagship model. ByteDance built this variant for rapid video drafting using multimodal inputs.
Seedance 2.5 Flash Pricing
The API costs $0.152 per generated second for 720p text-to-video or image-to-video output. Public documentation notes this is roughly 34 percent cheaper per second than the standard model. You must use the BytePlus ModelArk platform to access it.
Your account requires a minimum balance of $30.00, a resource pack, or an active savings plan to enable the model.
Speed and Performance Benchmarks
Vendor announcements state the Flash variant runs nearly twice as fast as the full Seedance 2.5 model. Speed is its primary design goal.
You should use this model for rapid drafts and multiple takes rather than final production renders. Independent benchmarks remain scarce, so treat vendor speed and quality claims with caution.
Core Technical Specifications
The model caps output resolution at 720p and 480p. It does not support 1080p or 4K resolutions.
| Feature | Seedance 2.5 Flash |
|---|---|
| Output Price | $0.152 per second |
| Input Token Price | $7.00 per million tokens |
| Max Duration | 30 seconds |
| Max Resolution | 720p |
| Image References | Up to 10 |
| Video References | Up to 5 |
| Audio References | Up to 5 |
Visual fidelity is lower than the full Seedance 2.5 model. Complex motion and intricate textures may appear less polished in the generated output.
Multimodal Reference Inputs
Prompts can include several types of reference material to guide generation. You can supply up to 10 reference images, 5 video clips, and 5 audio clips in a single request.
Video inputs are billed separately from output generation. The platform charges $7.00 per million tokens for 480p and 720p video inputs. The model also supports precise instruction-following and timestamp-level control during video editing tasks.
Output Duration and Limits
The system generates up to 30 seconds of continuous video in a single pass. Output includes both video and audio tracks.
The Flash tier is live right now inside the Dreamina app video tools, canvas, and agent features across supported regions in Europe, Asia, the Middle East, and South America. Developers can access the API via BytePlus ModelArk starting November 15, 2026, using the model ID dreamina-seedance-2-5-flash-261115.
Related on the GPUniq blog
Methodology
Facts are compiled from the vendor's own announcement and documentation, linked below, at the time of writing; where the vendor hasn't published a number it is marked as such. GPUniq prices, when the model is already available here, come from our live catalog.
Sources
- 1.One-take Creation, Flexible Referencing: Introducing Seedance 2.5 — ByteDance Seed Team (accessed )
- 2.Seed News – ByteDance Seed Team Blog — ByteDance / BytePlus (accessed )
- 3.Seedance 2.5 Flash: 30-second AI video, twice as fast — Senrama (accessed )
- 4.Dreamina Seedance 2.5 Flash announced by BytePlus — The Reel Stack (accessed )
- 5.Seedance 2.5 Flash: Price, Speed Claim and API Date — CellCog (accessed )
Want cheap GPUs for your next project?
Browse live GPU prices and rent the right card in seconds — H100, A100, RTX 4090, and 50+ more models.
