Model catalog
/v1/modelsNo authenticationThe response contains models, modes, parameter and media-input constraints, current availability, and Gensta customer prices in USD.
Variants and parameters
Each model declares its own variants and parameters. Use those values rather than assuming that every model has Standard and Prime, video resolutions, or the same defaults. The price and rules selected for a generation are captured when it is created.
Parameter defaults
Parameters below may be omitted: the API applies these defaults. Wan 3 requires an explicit variant (standard or prime) and a parameters object, which may be empty. Seedance defaults to variant standard; Seedream defaults to normal; their parameters object may be omitted. model, mode, and mode-specific input are required. Mode constraints take precedence: for example, edit/extend require ratio: adaptive. Defaults do not necessarily select the lowest price.
wan-3.0
| Parameter | Type | Values | Default |
|---|---|---|---|
| resolution | string | 480p, 720p, 1080p | "1080p" |
| duration_seconds | integer | 2–30 | 5 |
| ratio | string | adaptive, 16:9, 4:3, 1:1, 3:4, 9:16 | "adaptive" |
| audio | boolean | true, false | true |
| prompt_extend | boolean | true, false | false |
| watermark | boolean | true, false | false |
| num_outputs | integer | 1–1 | 1 |
| seed | integer | 0–2147483647 | omitted; selected by provider |
seedance-2.5
| Parameter | Type | Values | Default |
|---|---|---|---|
| resolution | string | 480p, 720p, 1080p | "720p" |
| duration_seconds | integer | 4–30; -1 (auto) | 5 |
| ratio | string | adaptive, 16:9, 4:3, 1:1, 3:4, 9:16, 21:9 | "adaptive" |
| audio | boolean | true, false | true |
| watermark | boolean | true, false | false |
| output_format | string | mp4 | "mp4" |
| return_last_frame | boolean | False, True | false |
| priority | integer | 0–9 | 0 |
| execution_expires_after_seconds | integer | 3600–259200 | 172800 |
| num_outputs | integer | 1–1 | 1 |
seedream-5.0-pro
| Parameter | Type | Values | Default |
|---|---|---|---|
| num_outputs | integer | 1–1 | 1 |
| size | string | 1K, 1.5K, 2K; WIDTHxHEIGHT | "1K" |
| prompt_optimization | string | standard, fast | "standard" |
| output_format | string | png, jpeg | "jpeg" |
| watermark | boolean | true, false | false |
seedream-5.0-pro · layer_decomposition
| Parameter | Type | Values | Default |
|---|---|---|---|
| size | string | auto, 1K, 1.5K, 2K | "auto" |
| output_format | string | png, jpeg | "png" |
| num_outputs | integer | 1 | 1 |
| watermark | boolean | true, false | false |
Billing
The catalog publishes customer-facing USD prices and a model-specific billing rule. Video models may charge successful output duration and verified video-input usage. Image models may charge by generated image, input count, or returned layers. The create response contains the maximum reserve; the terminal response contains the final charge and releases any unused reserve.
Mode constraints
operation_constraints contains the authoritative limits for each mode: required media slots, accepted kinds and formats, counts, sizes, durations, dimensions, ratios, and output artifacts. num_outputs always means alternative generation results—not input references or decomposition layers.