Models & Inputs
Video generation
Video models with per-length pricing and input fields.
POST /api/v1/processes with "type": "video-generation". Pick any model — the API does not plan-gate models. GET /api/v1/models returns the same capabilities at runtime.
Seedance generated audio
Seedance 2.x generates a native audio track by default. Set generateAudio: false for a silent video. When referenceAudios are attached, generated audio is required and generateAudio must be true or omitted.
{
"type": "video-generation",
"model": "seedance-2.5",
"input": {
"prompt": "A silent cinematic landscape reveal",
"videoLength": 5,
"generateAudio": false
}
}Genjutsu workflow
Use model: "genjutsu" with exactly one source video of 4 to 30 seconds in referenceVideos. Set genjutsuMode to "motion-transfer" (the default) to rebuild the scene using the source motion, or "object-swap" to replace a specified element. Add 1 to 8 referenceImages. Output can be 480p, 720p, or 1080p; duration and aspect ratio follow the source video.
model: "genjutsu-light" takes the same input at a lower price, for source videos of 4 to 15 seconds.
{
"type": "video-generation",
"model": "genjutsu",
"input": {
"prompt": "Replace the jacket with the one in the reference image",
"genjutsuMode": "object-swap",
"referenceVideos": [{ "assetId": "source_video_asset_id" }],
"referenceImages": [{ "assetId": "jacket_image_asset_id" }],
"resolution": 720
}
}Seedance 2.5 Edit workflow
Use model: "seedance-2.5-edit" for the dedicated base-video workflow. Put the base video first in referenceVideos; later video items are additional references. aspectRatio and durationMode are always "auto", and videoLength is not accepted.
editOperation defaults to "edit". Edit requests require a non-empty prompt; "extend", "sequel", and "prequel" work without one because the server supplies the complete continuation/lead-in instruction.
{
"type": "video-generation",
"model": "seedance-2.5-edit",
"input": {
"editOperation": "sequel",
"referenceVideos": [{ "assetId": "base_video_asset_id" }]
}
}Input fields
| Field | Type | Required | Description |
|---|---|---|---|
prompt | string | yes | What to generate. |
genjutsuMode | "motion-transfer" | "object-swap" | no | Only for genjutsu and genjutsu-light. Defaults to "motion-transfer"; use "object-swap" to replace a specific element in the source video. |
editOperation | "edit" | "extend" | "sequel" | "prequel" | no | Only for seedance-2.5-edit. Defaults to "edit"; prompt is optional for "extend", "sequel", and "prequel". |
aspectRatio | string | no | One of the model's aspect ratios, e.g. "16:9". Use "adaptive" when advertised to follow the primary image/video reference. |
resolution | number | no | Output resolution height, e.g. 1080. |
draft | boolean | no | Only for seedance-2.5 and seedance-2.5-edit. Every 480p run is a draft by default, billed like any 480p run; send false for a plain 480p render that cannot be finished. true sets resolution to 480, which must then stay 480. A completed draft reports draft.finishableUntil; finish it on model seedance-2.5 with draftVideoId, including edit drafts. |
draftVideoId | string | no | Only for seedance-2.5. The asset id of a completed draft from seedance-2.5 or seedance-2.5-edit (the draft process's output assetId). Renders the same shot, composition and motion at native 1080p, billed as a normal 1080p run of the draft's length and references. Send it alone: prompt, references, length, aspect ratio and audio all come from the draft, and sending any of them with draftVideoId is rejected. Must be used before the draft's finishableUntil (7 days after it was created). |
taskType | "auto" | "reference" | "edit" | "extend" | no | Only for seedance-2.5 and seedance-2.5-edit. Sent to BytePlus as omni_reference_task_type. Defaults to "auto", where Seedance decides from the prompt and references. "reference" generates a new video from the references; "edit" changes the reference video and needs aspectRatio and durationMode "auto"; "extend" continues it and needs aspectRatio "auto". Both need a referenceVideos item. Seedance still reads the prompt, and a prompt that reads as a different task type fails with TaskTypeMismatch. |
seed | integer | no | Only for seedance-2.5 and seedance-2.5-edit. Sent to BytePlus as seed, 0 to 2147483647. The same seed and inputs give a similar result, but Seedance does not guarantee an identical video. |
videoLength | number | no | Duration in seconds. |
startingFrameImageUrl | string | no | Image URL used as the first frame. External URLs are auto-imported as assets first. |
endingFrameImageUrl | string | no | Image URL used as the last frame. External URLs are auto-imported as assets first. |
durationMode | "fixed" | "auto" | no | Use "auto" to match the duration of a supported reference video; defaults to "fixed". |
generateAudio | boolean | no | Whether Seedance should generate a native audio track. Defaults to true. Must remain true when referenceAudios are attached. |
bitrateMode | "standard" | "high" | no | Seedance 2.x output bitrate preset. Use "high" for a crisper, larger file at the same credit cost; defaults to "standard". |
keyframes | {timeSeconds, imageUrl}[] | no | Timestamped image frames. timeSeconds is measured from the beginning of the generated clip. |
referenceImages | ({assetId} | {url})[] | no | Image references. Send either {assetId} for a Yapper asset or {url} for a public URL to import. Legacy {docId, url} objects remain accepted. |
referenceVideos | ({assetId} | {url})[] | no | Video references. Send either {assetId} for a Yapper asset or {url} for a public URL to import. Legacy {docId, url} objects remain accepted. |
referenceAudios | ({assetId} | {url})[] | no | Audio references. Send either {assetId} for a Yapper asset or {url} for a public URL to import. Legacy {docId, url} objects remain accepted; private storage metadata is resolved server-side. |
disableAutoRetries | boolean | no | Disable automatic same-model retries. Defaults to false, so automatic retries are enabled. |
customSettings | object | no | The model's own settings, the same ones the app's settings menu shows (for example GPT Image 2.5 quality and background: transparent). Only the settings listed in the model's capabilities.customSettings are accepted. Omitted settings keep their default. |
endingFrameImageRef | object | no | — |
outputName | string | no | — |
promptItemCount | number | no | — |
promptItemId | string | no | — |
promptItemIndex | number | no | — |
promptItemTitle | string | no | — |
startingFrameImageRef | object | no | — |
stripAndReattachAudio | boolean | no | — |
Referencing media: referenceImageUrls takes plain https image URLs. The normalized referenceImages, referenceVideos, and referenceAudios fields accept either {assetId} for an existing Yapper asset or {url} for public media, which is imported automatically. Provider-native asset:// URIs are internal and are not accepted here. Legacy {docId, url} objects remain accepted for backwards compatibility.
Example
Add "dryRun": true to get the exact credit cost without starting, and always send an Idempotency-Key header on real starts.
POST /api/v1/processes
{
"type": "video-generation",
"model": "heygen-video-1",
"input": {
"prompt": "A slow dolly shot over a foggy pine forest at sunrise",
"videoLength": 5,
"aspectRatio": "16:9"
}
}Models at a glance
Credits are per generated output; ranges span the supported resolutions/lengths (exact prices in each model's section).
| Model | Company | Credits | Lengths (s) | Max ref images | Frames | Audio |
|---|---|---|---|---|---|---|
genjutsu | higgsfield | dryRun | 4–30 | 8 | — | no |
genjutsu-light | higgsfield | dryRun | 4–15 | 8 | — | no |
wan-3.0-prime | alibaba | 120–1660 | 2–30 | 10 | start + end | yes |
wan-3.0 | alibaba | 100–1460 | 2–30 | 10 | start + end | yes |
flux-3 | black-forest-labs | 310–1240 | 5–20 | 0 | start + end | yes |
minimax-h3 | minimax | 50–160 | 5–15 | 9 | start + end | yes |
minimax-h3-max | minimax | 141–424 | 5–15 | 9 | start + end | yes |
minimax-h3-max-turbo | minimax | 71–212 | 5–15 | — | start + end | yes |
minimax-h3-singularity | minimax | 36–104 | 5–15 | 9 | start + end | yes |
heygen-video-1 | heygen | 13–38 | 5–15 | 9 | start | yes |
vidu-q4 | shengshu | 54–288 | 3–16 | 12 | start | yes |
seedance-2.5 | bytedance | 380–2800 | 4–30 | 30 | start + end | yes |
seedance-2.0 | bytedance | 220–840 | 4–15 | 9 | start + end | yes |
grok-imagine | xai | 58–135 | 4, 6, 8, 10, 12, 14, 15 | — | start | yes |
grok-imagine-v1.5 | xai | 20–260 | 1–15 | — | start | yes |
happy-horse | alibaba | 90–450 | 3–15 | 5 | start | yes |
seedance-2.0-fast | bytedance | 30–100 | 4–15 | 9 | start + end | yes |
seedance-2.0-mini | bytedance | 20–50 | 4–15 | 9 | start + end | yes |
wan-2.7 | alibaba | 30–150 | 2, 4, 6, 8, 10 | 5 | start + end | yes |
pixverse-v6 | pixverse | 20–370 | 1, 2, 4, 6, 8, 10, 12, 14, 15 | — | start | yes |
kling-4.0 | kuaishou | 100–1010 | 3, 5, 8, 10, 12, 15, 20, 25, 30 | 15 | start + end | yes |
kling-3.0 | kuaishou | 72–270 | 4, 6, 8, 10, 12, 14, 15 | 3 | start + end | yes |
kling-3.0-pro | kuaishou | 96–360 | 4, 6, 8, 10, 12, 14, 15 | 3 | start + end | yes |
gemini-omni-flash | 115–381 | 3–10 | 10 | start + end | yes | |
veo3-quality | 216 | 8 | 3 | start + end | yes | |
veo3-fast | 80 | 8 | 3 | start + end | yes | |
seedance-2.0-open | bytedance | 310–1180 | 4–15 | 9 | start + end | yes |
seedance-2.5-edit | bytedance | dryRun | — | 30 | — | yes |
Genjutsu (genjutsu)
Recast a source video's motion or swap a specific element using reference images.
| Credits | Cost depends on the source video duration and output resolution. POST /processes with dryRun: true for an exact quote. |
| Aspect ratios | auto — referenceVideos require "auto" (or omission) |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 |
| Resolutions | 480, 720, 1080 |
| Reference images | up to 8 |
| Reference videos | up to 1 |
Genjutsu Light (genjutsu-light)
A lower-cost Genjutsu for source clips up to 15 seconds: recast the motion or swap a specific element using reference images.
| Credits | Cost depends on the source video duration and output resolution. POST /processes with dryRun: true for an exact quote. |
| Aspect ratios | auto — referenceVideos require "auto" (or omission) |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720, 1080 |
| Reference images | up to 8 |
| Reference videos | up to 1 |
WAN 3.0 Prime (wan-3.0-prime)
Alibaba's high-speed WAN 3.0 tier, around 7x faster with the same multimodal inputs and native audio.
| Credits | 2s: 120 · 3s: 170 · 4s: 230 · 5s: 280 · 6s: 340 · 7s: 390 · 8s: 450 · 9s: 500 · 10s: 560 · 11s: 610 · 12s: 670 · 13s: 720 · 14s: 780 · 15s: 830 · 16s: 890 · 17s: 940 · 18s: 1000 · 19s: 1050 · 20s: 1110 · 21s: 1160 · 22s: 1220 · 23s: 1270 · 24s: 1330 · 25s: 1390 · 26s: 1440 · 27s: 1500 · 28s: 1550 · 29s: 1610 · 30s: 1660 (at 1080p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 |
| Resolutions | 480, 720, 1080 |
customSettings | generate_audio_switch: true or false (default true)enable_thinking: true or false (default false) |
| Reference images | up to 10 |
| Reference videos | up to 5 (15s total) |
| Reference audio | up to 5 (15s total) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 5000 |
WAN 3.0 (wan-3.0)
Alibaba's newest multimodal video model with native audio, start/end frames, and image, video, and audio references.
| Credits | 2s: 100 · 3s: 150 · 4s: 200 · 5s: 250 · 6s: 300 · 7s: 350 · 8s: 390 · 9s: 440 · 10s: 490 · 11s: 540 · 12s: 590 · 13s: 640 · 14s: 690 · 15s: 730 · 16s: 780 · 17s: 830 · 18s: 880 · 19s: 930 · 20s: 980 · 21s: 1030 · 22s: 1070 · 23s: 1120 · 24s: 1170 · 25s: 1220 · 26s: 1270 · 27s: 1320 · 28s: 1370 · 29s: 1410 · 30s: 1460 (at 1080p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 |
| Resolutions | 480, 720, 1080 |
customSettings | generate_audio_switch: true or false (default true)enable_thinking: true or false (default false) |
| Reference images | up to 10 |
| Reference videos | up to 5 (15s total) |
| Reference audio | up to 5 (15s total) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 5000 |
FLUX.3 (flux-3)
Black Forest Labs' frontier video model for text, timestamped image keyframes, start/end-frame interpolation, or video continuation, with native audio and clips up to 20 seconds.
| Credits | 5s: 310 · 6s: 370 · 7s: 440 · 8s: 500 · 9s: 560 · 10s: 620 · 11s: 680 · 12s: 750 · 13s: 810 · 14s: 870 · 15s: 930 · 16s: 990 · 17s: 1060 · 18s: 1120 · 19s: 1180 · 20s: 1240 (at 1080p) |
| Aspect ratios | 16:9, 9:16, 1:1, 21:9, 2:1, 4:3, 3:4, auto (default auto) |
| Lengths (s) | 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 |
| Resolutions | 720, 1080, 1440, 2160 |
customSettings | generate_audio_switch: true or false (default true) |
| Reference images | up to 0 |
| Reference videos | up to 1 |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
MiniMax H3 (minimax-h3)
MiniMax's frontier 2K video model for text generation and multimodal image, video, and audio references.
| Credits | 5s: 50 · 6s: 70 · 7s: 80 · 8s: 90 · 9s: 100 · 10s: 110 · 11s: 120 · 12s: 130 · 13s: 140 · 14s: 150 · 15s: 160 (at 768p) |
| Aspect ratios | 16:9, 21:9, 4:3, 1:1, 3:4, 9:16, adaptive |
| Lengths (s) | 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 768, 1440 |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 50000 |
MiniMax H3 Max (minimax-h3-max)
Fal's post-trained H3 variant for exceptionally fast 480p, 768p and 1080p video from text, first/end-frame animation, or image, video, and audio references, with native audio.
| Credits | 5s: 141 · 6s: 169 · 7s: 198 · 8s: 226 · 9s: 254 · 10s: 282 · 11s: 311 · 12s: 339 · 13s: 367 · 14s: 395 · 15s: 424 (at 1080p) |
| Aspect ratios | 16:9, 21:9, 4:3, 1:1, 3:4, 9:16 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 768, 1080 |
customSettings | prompt_expansion_mode: disabled, balanced, quality (default balanced) |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
MiniMax H3 Max Turbo (minimax-h3-max-turbo)
The fastest and cheapest H3 variant, generating 480p, 768p and 1080p text-to-video or first/end-frame animation with native audio in roughly half the time of H3 Max.
| Credits | 5s: 71 · 6s: 85 · 7s: 99 · 8s: 113 · 9s: 127 · 10s: 141 · 11s: 155 · 12s: 169 · 13s: 184 · 14s: 198 · 15s: 212 (at 1080p) |
| Aspect ratios | 16:9, 21:9, 4:3, 1:1, 3:4, 9:16 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 768, 1080 |
customSettings | prompt_expansion_mode: disabled, balanced, quality (default balanced) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
MiniMax H3 Singularity (minimax-h3-singularity)
Community-tuned H3 with native audio. Choose Economy for a lower price or Fast for faster processing, with the same output settings.
| Credits | 5s: 36 · 6s: 46 · 7s: 51 · 8s: 55 · 9s: 65 · 10s: 70 · 11s: 80 · 12s: 85 · 13s: 94 · 14s: 99 · 15s: 104 (at 1080p) |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4, 3:2, 2:3, 21:9 |
| Lengths (s) | 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720, 832, 1080, 1440, 2160 |
customSettings | singularity_speed: economy, fast (default fast)singularity_seed: -1 to 281474976710655 (default -1)singularity_first_steps: 1 to 64 (default 8)singularity_refine_steps: 0 to 64 (default 4)singularity_upscale: 1 to 4 (default 1.5)singularity_base_megapixels: 0 to 16 (default 0)singularity_denoise: 0.01 to 1 (default 1)singularity_lora_strength: 0 to 2 (default 1)singularity_sampler: euler, heun, dpmpp_2m, dpmpp_2m_sde (default euler)singularity_scheduler: beta, normal, simple, karras, exponential, ddim_uniform (default beta)singularity_reference_size: max, match (default max)singularity_output_audio: true or false (default true)singularity_reference_video_audio: true or false (default false)singularity_crf: 0 to 51 (default 19) |
| Reference images | up to 9 |
| Reference videos | up to 3 (45s total) |
| Reference audio | up to 3 (45s total) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 20000 |
HeyGen Video (heygen-video-1)
HeyGen's post-trained H3 model for fast, low-cost 480p and 768p video from text, a first frame, or image, video, and audio references, with native audio.
| Credits | 5s: 13 · 6s: 15 · 7s: 18 · 8s: 20 · 9s: 23 · 10s: 25 · 11s: 28 · 12s: 30 · 13s: 33 · 14s: 35 · 15s: 38 (at 480p) |
| Pricing | Limited-time launch price; the future standard credit price is 2× the current rate |
| Aspect ratios | 16:9, 21:9, 4:3, 1:1, 3:4, 9:16 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 768 |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | supported |
| Start/end frames | start |
| Audio output | yes |
Vidu Q4 (vidu-q4)
Vidu's reference-driven video model: keep up to 12 image references and 3 voices consistent, or start from text or a first frame. Native audio, 3 to 16 seconds, up to 4K.
| Credits | 3s: 54 · 4s: 72 · 5s: 90 · 6s: 108 · 7s: 126 · 8s: 144 · 9s: 162 · 10s: 180 · 11s: 198 · 12s: 216 · 13s: 234 · 14s: 252 · 15s: 270 · 16s: 288 (at 1080p) |
| Pricing | Limited-time launch price; the future standard credit price is 1.43× the current rate |
| Aspect ratios | 16:9, 9:16, 4:3, 3:4, 1:1 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 |
| Resolutions | 540, 720, 1080, 1440, 2160 |
| Reference images | up to 12 |
| Reference audio | up to 3 |
| Text-only generation | supported |
| Start/end frames | start |
| Audio output | yes |
| Max prompt length | 5000 |
Seedance 2.5 (seedance-2.5)
ByteDance's latest multimodal video model with native audio, 30-second generations, and expanded image, video, and audio references.
| Credits | 4s: 380 · 5s: 470 · 6s: 570 · 7s: 660 · 8s: 750 · 9s: 850 · 10s: 940 · 11s: 1030 · 12s: 1130 · 13s: 1220 · 14s: 1310 · 15s: 1400 · 16s: 1500 · 17s: 1590 · 18s: 1680 · 19s: 1780 · 20s: 1870 · 21s: 1960 · 22s: 2060 · 23s: 2150 · 24s: 2240 · 25s: 2340 · 26s: 2430 · 27s: 2520 · 28s: 2610 · 29s: 2710 · 30s: 2800 (at 1080p) |
| Pricing | Limited-time launch price; the future standard credit price is 1.5× the current rate |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 |
| Resolutions | 480, 720, 1080 |
| Bitrate modes | standard, high (default standard) |
| Generated audio | configurable (default on; required with reference audio) |
| Draft mode | 480p runs are drafts unless draft: false; finish the same shot at 1080p with draftVideoId on seedance-2.5 within 7 days |
| Reference images | up to 30 |
| Reference videos | up to 10 (30s total) |
| Reference audio | up to 10 (30s total) |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 20000 |
Seedance 2.0 (seedance-2.0)
ByteDance's video model with multi-modal input, lip sync, and multi-shot narrative.
| Credits | 4s: 220 · 5s: 280 · 6s: 340 · 7s: 390 · 8s: 450 · 9s: 510 · 10s: 560 · 11s: 620 · 12s: 670 · 13s: 730 · 14s: 790 · 15s: 840 (at 1080p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720, 1080, 2160 |
| Bitrate modes | standard, high (default standard) |
| Generated audio | configurable (default on; required with reference audio) |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | not supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 20000 |
Grok Imagine (grok-imagine)
Grok's best and latest video model with sound included.
| Credits | 4s: 58 · 6s: 72 · 8s: 86 · 10s: 100 · 12s: 114 · 14s: 128 · 15s: 135 (at 720p) |
| Aspect ratios | 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 1:1 — "auto" also accepted with startingFrameImageUrl |
| Lengths (s) | 4, 6, 8, 10, 12, 14, 15 |
| Resolutions | 720 |
| Text-only generation | supported |
| Start/end frames | start |
| Audio output | yes |
| Max prompt length | 4096 |
Grok Imagine 1.5 (grok-imagine-v1.5)
xAI's Grok Imagine 1.5 image-to-video model with audio included.
| Credits | 1s: 20 · 2s: 40 · 3s: 50 · 4s: 70 · 5s: 90 · 6s: 110 · 7s: 120 · 8s: 140 · 9s: 160 · 10s: 170 · 11s: 190 · 12s: 210 · 13s: 230 · 14s: 240 · 15s: 260 (at 480p) |
| Aspect ratios | 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 1:1 — "auto" also accepted with startingFrameImageUrl |
| Lengths (s) | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720 |
| Text-only generation | not supported |
| Start/end frames | start |
| Audio output | yes |
| Max prompt length | 4096 |
Happy Horse (happy-horse)
Alibaba's reference-to-video model with native audio and 720p output.
| Credits | 3s: 90 · 4s: 120 · 5s: 150 · 6s: 180 · 7s: 210 · 8s: 240 · 9s: 270 · 10s: 300 · 11s: 330 · 12s: 360 · 13s: 390 · 14s: 420 · 15s: 450 (at 720p) |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Lengths (s) | 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 720 |
| Reference images | up to 5 |
| Text-only generation | not supported |
| Start/end frames | start |
| Audio output | yes |
| Max prompt length | 5000 |
Seedance 2.0 Fast (seedance-2.0-fast)
Fast variant of ByteDance's video model with multi-modal input, lip sync, and multi-shot narrative.
| Credits | 4s: 30 · 5s: 40 · 6s: 40 · 7s: 50 · 8s: 50 · 9s: 60 · 10s: 70 · 11s: 70 · 12s: 80 · 13s: 80 · 14s: 90 · 15s: 100 (at 480p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720 |
| Bitrate modes | standard, high (default standard) |
| Generated audio | configurable (default on; required with reference audio) |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | not supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 20000 |
Seedance 2.0 Mini (seedance-2.0-mini)
ByteDance's economical video model for fast multimodal drafts with native audio.
| Credits | 4s: 20 · 5s: 20 · 6s: 20 · 7s: 30 · 8s: 30 · 9s: 30 · 10s: 40 · 11s: 40 · 12s: 40 · 13s: 40 · 14s: 50 · 15s: 50 (at 480p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720 |
| Bitrate modes | standard, high (default standard) |
| Generated audio | configurable (default on; required with reference audio) |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | not supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 20000 |
WAN 2.7 (wan-2.7)
Wan's latest reference-driven video model with image and video guidance for better consistency.
| Credits | 2s: 30 · 4s: 60 · 6s: 90 · 8s: 120 · 10s: 150 (at 1080p) |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Lengths (s) | 2, 4, 6, 8, 10 |
| Resolutions | 720, 1080 |
| Reference images | up to 5 |
| Reference videos | up to 1 |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 5000 |
PixVerse V6 (pixverse-v6)
PixVerse's latest Fal-hosted video model with optional audio and first-frame image support.
| Credits | 1s: 20 · 2s: 50 · 4s: 100 · 6s: 150 · 8s: 200 · 10s: 250 · 12s: 300 · 14s: 350 · 15s: 370 (at 1080p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16, 2:3, 3:2, 21:9 — startingFrameImageUrl requires "auto" (or omission) |
| Lengths (s) | 1, 2, 4, 6, 8, 10, 12, 14, 15 |
| Resolutions | 360, 540, 720, 1080 |
customSettings | generate_audio_switch: true or false (default true)generate_multi_clip_switch: true or false (default true) |
| Text-only generation | supported |
| Start/end frames | start |
| Audio output | yes |
| Max prompt length | 2048 |
Kling 4.0 (kling-4.0)
Kling's newest video model: up to 30 seconds at 1080p with native audio, start/end frames, image and element references, and video editing.
| Credits | 3s: 100 · 5s: 170 · 8s: 270 · 10s: 340 · 12s: 400 · 15s: 500 · 20s: 670 · 25s: 840 · 30s: 1010 (at 1080p) |
| Aspect ratios | 16:9, 9:16, 1:1, 21:9 — referenceVideos require "auto" (or omission) |
| Lengths (s) | 3, 5, 8, 10, 12, 15, 20, 25, 30 |
| Resolutions | 720, 1080 |
| Reference images | up to 15 |
| Reference videos | up to 1 |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 2500 |
Kling 3.0 (kling-3.0)
Video model with start/end frames, references, and native audio.
| Credits | 4s: 72 · 6s: 108 · 8s: 144 · 10s: 180 · 12s: 216 · 14s: 252 · 15s: 270 (at 720p) |
| Aspect ratios | 16:9, 9:16, 1:1 — startingFrameImageUrl and referenceVideos require "auto" (or omission) |
| Lengths (s) | 4, 6, 8, 10, 12, 14, 15 |
| Resolutions | 720 |
customSettings | keepAudio: true or false (default true) |
| Reference images | up to 3 |
| Reference videos | up to 1 |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 2500 |
Kling 3.0 Pro (kling-3.0-pro)
Higher-cost Kling model with start/end frame, reference, and native audio support.
| Credits | 4s: 96 · 6s: 144 · 8s: 192 · 10s: 240 · 12s: 288 · 14s: 336 · 15s: 360 (at 1080p) |
| Aspect ratios | 16:9, 9:16, 1:1 — startingFrameImageUrl and referenceVideos require "auto" (or omission) |
| Lengths (s) | 4, 6, 8, 10, 12, 14, 15 |
| Resolutions | 1080 |
customSettings | keepAudio: true or false (default true) |
| Reference images | up to 3 |
| Reference videos | up to 1 |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 2500 |
Gemini Omni 1.1 Flash (gemini-omni-flash)
Google's fast multimodal video model with native audio, first/end-frame interpolation, video editing and extension, and output up to 4K.
| Credits | 3s: 115 · 4s: 153 · 5s: 191 · 6s: 229 · 7s: 267 · 8s: 305 · 9s: 343 · 10s: 381 (at 1080p) |
| Aspect ratios | 16:9, 9:16 — referenceVideos require "auto" (or omission) |
| Lengths (s) | 3, 4, 5, 6, 7, 8, 9, 10 |
| Resolutions | 360, 720, 1080, 2160 |
| Reference images | up to 10 |
| Reference videos | up to 1 |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
Veo3.1 Quality (veo3-quality)
Google's highest quality video model with sound included. Blocks celebrities.
| Credits | 8s: 216 (at 1080p) |
| Aspect ratios | 16:9, 9:16 |
| Lengths (s) | 8 |
| Resolutions | 1080 |
| Reference images | up to 3 |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
Veo3.1 (veo3-fast)
Google's latest video model with sound included. Blocks celebrities.
| Credits | 8s: 80 (at 1080p) |
| Aspect ratios | 16:9, 9:16 |
| Lengths (s) | 8 |
| Resolutions | 1080 |
| Reference images | up to 3 |
| Text-only generation | supported |
| Start/end frames | start + end |
| Audio output | yes |
Seedance 2.0 Open (seedance-2.0-open)
Less restricted Seedance 2.0 variant for content that the standard model may block.
| Credits | 4s: 310 · 5s: 390 · 6s: 480 · 7s: 550 · 8s: 630 · 9s: 710 · 10s: 780 · 11s: 870 · 12s: 940 · 13s: 1020 · 14s: 1110 · 15s: 1180 (at 1080p) |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive |
| Lengths (s) | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 |
| Resolutions | 480, 720, 1080, 2160 |
| Bitrate modes | standard, high (default standard) |
| Generated audio | configurable (default on; required with reference audio) |
| Reference images | up to 9 |
| Reference videos | up to 3 (15s total) |
| Reference audio | up to 3 (15s total) |
| Text-only generation | not supported |
| Start/end frames | start + end |
| Audio output | yes |
| Max prompt length | 20000 |
Seedance 2.5 Edit (seedance-2.5-edit)
Dedicated Seedance 2.5 base-video workflow for editing or extending a video, or generating a natural sequel or prequel. The first referenceVideos item is the base video; duration and aspect ratio are automatic. Prompt is optional for extend, sequel, and prequel, but required for edit.
| Credits | Cost depends on the base video's automatic output duration — POST /processes with dryRun: true for an exact quote. |
| Pricing | Limited-time launch price; the future standard credit price is 1.5× the current rate |
| Base video | required as the first referenceVideos item (array index 0) |
| Operations | edit, extend, sequel, prequel |
| Prompt | required for edit; optional for extend and sequel and prequel |
| Duration | automatic (durationMode: "auto") |
| Aspect ratios | auto (default auto) — referenceVideos require "auto" (or omission) |
| Resolutions | 480, 720, 1080 |
| Bitrate modes | standard, high (default standard) |
| Generated audio | configurable (default on; required with reference audio) |
| Draft mode | 480p runs are drafts unless draft: false; finish the same shot at 1080p with draftVideoId on seedance-2.5 within 7 days |
| Reference images | up to 30 |
| Reference videos | up to 10 (30s total) |
| Reference audio | up to 10 (30s total) |
| Text-only generation | not supported |
| Audio output | yes |
| Max prompt length | 20000 |