Can Wan 3.0 run in ComfyUI?
Yes. Official API Nodes and templates support text, image and reference-to-video.
Verified Sep 14, 2026
Wan 3.0 in ComfyUI
Generate with Wan 3.0 from ComfyUI using hosted API Nodes. Your workflow can run on your computer; video generation runs in the cloud.
3API templates
v0.33.4+minimum ComfyUI version
30smaximum combined video duration
—local model weights
InputText · image · video · audio · file · link
One modelwan3.0-video
OutputUp to 30s · 1080P · native audio
Can Wan 3.0 run locally?
No verified official weights or runnable local inference release were found as of September 14, 2026.
Does ComfyUI download Wan 3.0 weights?
No. These API Nodes submit generation requests to a hosted service.
Get started
- Update ComfyUI to v0.33.4 or newer, or open Comfy Cloud.
- Sign in to ComfyUI and check your credits. API Nodes use Comfy billing; an Alibaba Model Studio free quota does not automatically apply.
- Open Templates and choose one of the three Wan 3.0 workflows below. Add your prompt and references, review the cost, then run.
Three official workflow templates
Original examples for adaptation; these prompts are not presented as tested generation results.
Wan 3.0 Text to Video
Text to Video
Start from a written scene. Describe one subject, its motion, camera direction and sound.
wan3.0-video480P · 8s · 16:9Reference setup
No reference assets are required.
Example prompt
An 8-second continuous shot of a ceramic cup on a sunlit wooden table. The camera slowly pushes in as steam rises from the coffee. Soft morning light, natural textures, quiet room ambience and a gentle cup clink. No dialogue or music.Wan 3.0 Image to Video
Image to Video
Load a first-frame image. Describe what changes while keeping the original subject and framing.
wan3.0-video480P · 8s · 16:9Reference setup
Supply one first-frame image in the image input.
Example prompt
Use the supplied image as the first frame. Over 8 seconds, move the camera slowly forward while a light breeze moves the fabric. Preserve the subject, clothing, background and lighting from the first frame. Natural movement, soft ambient wind, no dialogue.Wan 3.0 Reference to Video
Reference to Video
Attach image, video and audio references, then mention each asset by its numbered tag.
wan3.0-video480P · 8s · 16:9Reference setup
For this example, attach a character image as @Image1, a 5-second camera-reference video as @Video1 and a voice reference as @Audio1.
Example prompt
Create an 8-second shot. Use @Image1 for the character appearance, @Video1 for the slow camera movement, and @Audio1 for the voice timbre. The character turns toward the camera and says, "Let us begin." Preserve facial identity and clothing, with natural lip sync and quiet room ambience.Reference syntax and duration
Use the documented tag spelling and number assets within each media type: @Image1, @Video1, @Audio1. They point to assets attached to the reference node; typing a tag alone does not upload a file.
With reference video, input-video seconds plus output seconds must stay within 30 seconds. A 5-second input and an 8-second output total 13 seconds, and both sides are billable.
Costs and local-run status
API Nodes require online access and credits. Check the price shown in ComfyUI before running; direct Alibaba Model Studio pricing is a separate access path. There is no Wan 3.0 checkpoint to download from these templates.
Missing nodes?
Update ComfyUI and reload its interface. If a template shows an unknown node, check the official announcement and API Nodes requirements linked below.