Wan 3.0 Prime (BudgetPixel)

Wan 3.0 Prime — the faster tier of the all-in-one, for when patience costs more than credits

By budgetpixel-ai·Created about 17 hours ago·Updated about 12 hours ago· 3

end_image
image
reference_audios
reference_images
reference_videos

video

◄prompt►

◄aspect_ratioadaptive►

◄generate_audiotrue►

◄length_seconds5►

◄resolution720p►

◄seed-1►

◄force_rerunfalse►

Wan 3.0 Prime is the "pay more, wait less" tier of Alibaba's Wan 3.0 video model - 170 credits per output second against the standard 3.0's 120, in exchange for significantly faster end-to-end generation. Everything else is identical: the same all-in-one modes (text, image with first/last frame, and reference-to-video with up to 10 images, 5 video clips, and 5 audio clips), the same 2–30 second output window, the same 480p/720p/1080p choices, and the same refreshing rule that only output seconds are billed - input media stays free.

When to pay the premium

The decision is a time-vs-money one, and it's worth doing the math before defaulting to Prime. A draft or an iterated prompt where you'll regenerate ten times is a bad Prime use - that 40-credit-per-second premium (170 vs 120) compounds fast over a run of tries. Where Prime earns it is the deadline job: a long clip that needs to come back in minutes, or a pipeline where the polling time is the bottleneck. If your workflow is one shot and done, and your time is worth something, the faster return is real.

The inputs that matter

Identical to Wan 3.0 - the schema doesn't drop anything in the faster tier:

prompt - required; edit instruction with reference videos; assets addressable by order ("Image 1", "Video 1", "Audio 1").
reference_images - up to 10, free.
reference_videos - up to 5 clips (15s combined); input duration + length_seconds ≤ 30s.
reference_audios - up to 5 clips (15s combined), free.
image / end_image - first/last frames; can't combine with reference media.
length_seconds - 2–30.
resolution - 480p/720p/1080p.
aspect_ratio - adaptive default.
generate_audio - defaults true, no price impact.
seed - -1 random; set to reproduce.

Output is video (VIDEO), core type - 2025+ ComfyUI.

How it works

Same hosted mechanics as the standard node - serialize, upload, create a job, poll, download the MP4. The only difference is a shorter server-side turnaround. Interrupting the graph still leaves the job running and billing.

Install and key setup

ComfyUI Manager → BudgetPixel, or:

cd ComfyUI/custom_nodes
git clone https://github.com/budgetpixel-ai/comfyui-budgetpixel.git
pip install -r comfyui-budgetpixel/requirements.txt

Only requests. API key from a paid plan via BUDGETPIXEL_API_KEY or budgetpixel.json; never in workflow JSON.

Where people get burned

The premium is the gotcha - it's easy to set-and-forget Prime and pay 40 credits/second extra on every draft. Be deliberate: Prime for final takes and deadlines, standard Wan 3.0 for iteration. The 30-second combined cap and reference-versus-frames exclusivity apply exactly as in the standard node, and ComfyUI's output cache means force_rerun is your friend when you want a fresh result.

CategoryBudgetPixel/Video

Inputs (12)

Name	Type	Default	Description
prompt	STRING		Text description of the video, or the edit instruction when reference videos are supplied. In reference mode, address assets by order: "Image 1", "Video 1", "Audio 1".
aspect_ratioopt	COMBO	adaptive	Aspect ratio, honored in every mode (an explicit ratio reframes the output even with a first frame or reference media). "adaptive" (default) lets the model pick a suitable ratio from the inputs and prompt intent.
end_imageopt	IMAGE		Optional last frame, used together with `image` (the first frame) to interpolate the video between the two frames. Same input forms as `image`.
generate_audioopt	BOOLEAN	true	Generate audio with the video (default true; no price impact).
imageopt	IMAGE		Optional first frame for image-to-video. Provide a public image URL, a data URI, raw base64, or an uploaded-file URL from POST /v1/uploads. Omit for text-to-video. Can't be combined with reference media.
length_secondsopt	INT	52–30	Output video length in seconds (2-30). With reference videos, input duration + output length must not exceed 30 seconds.
reference_audiosopt	AUDIO		Optional reference audio clips (up to 5; each 2-15s, 15s combined; WAV/MP3) that guide sound/voice. Free. Each item is a public audio URL or an uploaded-file URL from POST /v1/uploads. Can't be combined with `image`/`end_image`.
reference_imagesopt	IMAGE		Optional reference images (up to 10, free) that guide identity/style/scene in reference-to-video mode. Each item is a public image URL, a data URI, raw base64, or an uploaded-file URL from POST /v1/uploads. Can't be combined with `image`/`end_image`.
reference_videosopt	VIDEO		Optional reference video clips (up to 5; each 2-15s, 15s combined; MP4/MOV, ≤50MB each) for editing, effect/camera replication, and extension. Each item is a public video URL or an uploaded-file URL from POST /v1/uploads (videos are passed by URL, not inlined). Free — only OUTPUT seconds are billed — but input video duration + `length_seconds` must not exceed 30. Can't be combined with `image`/`end_image`.
resolutionopt	COMBO	720p	Output resolution. Pricing varies by resolution — see `resolution_pricing` in GET /v1/models.
seedopt	INT	-1-1–2147483647	Seed for reproducible generation. Omit for random.
force_rerunopt	BOOLEAN	false	—

Outputs (1)

Name	Type	Description
video	VIDEO	—