Skip to main content

Video Generation

We offer two video generation model series — Gemini Veo and Doubao Seedance — covering a wide range of scenarios including text-to-video, image-to-video, video continuation, and audio-driven generation.

Icon Legend

Input: Text · Image · Video
Output: Video

Gemini Veo

ProviderModel IDCapabilitiesEndpointSupported DurationPrice (USD/sec)Launch DateExpected DeprecationCapacitySupported RegionsNotes
Vertex AIturing/veo-3.1-generateInput:
Output:
v1/videos"4" / "6" / "8"$0.40--Vertex AI multi-projectChina region
Europe region
North America region
High-quality version, recommended for final production output
Vertex AIturing/veo-3.1-fast-generateInput:
Output:
v1/videos"4" / "6" / "8"$0.15--Vertex AI multi-projectChina region
Europe region
North America region
Low-latency version, suitable for rapid iteration

Pricing Examples

  • veo-3.1-generate: 6s × $0.40 = $2.40
  • veo-3.1-fast-generate: 6s × $0.15 = $0.90

Doubao Seedance

ProviderModel IDCapabilitiesEndpointSupported DurationBillingLaunch DateExpected DeprecationCapacitySupported RegionsNotes
Volcano Enginedoubao-seedance-2-0-260128Input:
Output:
v1/videos"4" / "5" / "6" / "8"¥46/M tokens (video input ¥28/M)--Volcano Engine ArkChina regionHigh-quality version; supports text-to-video, image-to-video, video continuation, and audio-driven generation
Volcano Enginedoubao-seedance-2-0-fast-260128Input:
Output:
v1/videos"4" / "5" / "6" / "8"¥37/M tokens (video input ¥22/M)--Volcano Engine ArkChina regionLow-latency version, suitable for rapid iteration
Volcano Enginedoubao-seedance-2-5-260628Input:
Output:
v1/videos"4" – "30"480p/720p ¥70/M tokens (video input ¥42/M)
1080p ¥77/M tokens (video input ¥46/M)
--Volcano Engine ArkChina regionNative 1080p output and up to 30 seconds in a single clip

Supported Resolutions

Seedance 2.0 / 2.0 fast / 2.5 support the following resolutions (passed via the size parameter, automatically mapped to resolution + ratio). The two model generations render different 480p 16:9 / 9:16 dimensions and the platform accepts either spelling. 1080p is served by 2.0 and 2.5 only, 4k by 2.0 only, and 2.0 fast is limited to 480p / 720p:

ResolutionAspect RatioSeedance 2.5Seedance 2.0 series
480p16:9854×480864×496
480p4:3752×560752×560
480p1:1640×640640×640
480p3:4560×752560×752
480p9:16480×854496×864
480p21:9992×432992×432
720p16:91280×7201280×720
720p4:31112×8341112×834
720p1:1960×960960×960
720p3:4834×1112834×1112
720p9:16720×1280720×1280
720p21:91470×6301470×630
1080p16:91920×10801920×1080
1080p4:31664×12481664×1248
1080p1:11440×14401440×1440
1080p3:41248×16641248×1664
1080p9:161080×19201080×1920
1080p21:92206×9462206×946
4k16:93840×2160
4k4:33326×2494
4k1:12880×2880
4k3:42494×3326
4k9:162160×3840
4k21:94398×1886

For video editing, video extension and first-frame / first-and-last-frame tasks, Seedance 2.5 accepts ratio: "adaptive" only — do not use size in those scenes (it derives a concrete aspect ratio); pass resolution instead.

On 2.5, ratio accepts 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 and adaptive (the model decides), and duration ranges from 4 to 30 seconds.

Pricing Details

Seedance uses token-based billing in CNY, unlike Veo's per-second pricing:

  • doubao-seedance-2-0-260128: ¥46 / million tokens (¥28 / million tokens when video input is included)
  • doubao-seedance-2-0-fast-260128: ¥37 / million tokens (¥22 / million tokens when video input is included)
  • doubao-seedance-2-5-260628: 480p/720p ¥70 / million tokens (¥42 with video input); 1080p ¥77 / million tokens (¥46 with video input)