All articles
Image & Video Generation APIs/

Doubao Seedance 2.5 Flash API Guide

Official pricing, speed benchmarks, and specs for ByteDance's video API.

2 minutes read

Doubao Seedance 2.5 Flash API Guide. A macro shot of a dense server rack illuminated by amber status LEDs, where meticulously routed fiber-optic cables cast sharp reflections across polished black silicon heatsinks — GPUniq
Data as of 3 sources

Drafted with AI tools and edited by a human. The figures and conclusions were checked by a GPUniq editor.

TL;DR

The Doubao Seedance 2.5 Flash API costs $7.00 per 1M tokens for 480p and 720p generations without video input, dropping to $4.30 per 1M tokens with reference video inputs. It generates video 2 to 3 times faster than the standard Seedance 2.5 model, bringing 720p costs to roughly $0.15 per second while capping clips at 30 seconds.

API Pricing

Base pricing sits at $7.00 per 1M tokens for 480p and 720p generations without video input. Supplying reference video inputs drops the rate to $4.30 per 1M tokens.

The effective price at 720p works out to approximately $0.15 per second, compared to approximately $0.23 per second for the standard Seedance 2.5 model.

Feature TierPrice, USD per 1M tokensEffective Price, USD/sec
Standard (No Video Input)7.00N/A
Discounted (Reference Video Input)4.30N/A

Benchmarks and Speeds

Generation speed runs roughly 2 to 3 times faster than the full Seedance 2.5 model at the same resolution.

ByteDance released the model on November 15, 2026, under model IDs doubao-seedance-2-5-flash-261115 and dreamina-seedance-2-5-flash-261115.

Resolution Limits

The model outputs native 480p and 720p resolutions. It does not support 1080p or 4K generation.

Quality can dip below the standard Seedance 2.5 model when handling complex scenes or high-resolution demands.

Inputs and Capabilities

The API accepts text-to-video, reference images, reference videos, audio references, and first or last frame constraints. It also generates native audio alongside the video.

Reference video inputs trigger the lower $4.30 per 1M token tier.

Max Clip Length

Generated clips run from 4 to 30 seconds per request.

Access and Availability

ByteDance offers the model internationally via BytePlus ModelArk and domestically in China via Volcengine Ark.

Aggregators and third-party catalogs like WaveSpeed, BudgetPixel, AnyAIGC, and Atlas Cloud also list the endpoint. Offline inference is not supported.

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. 1.One-take Creation, Flexible Referencing: Introducing Seedance 2.5 — ByteDance Seed Team (accessed )
  2. 2.ByteDance Seedance 2.5 Flash API: 720p Pricing, Speed & ModelArk Setup — LLM API Guide (accessed )
  3. 3.Doubao Seedance 2.5 Flash (261115) pricing, context and providers — OhMyGPT — OhMyGPT / Models.dev (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.

More from the GPUniq blog