ComfyUI Razv WaveSpeed Nodes

Instruction-based image editing, hosted — Wan 2.5's edit model on a URL

Seedream 4.0 at up to 4096×4096, without hunting down an API script

Seedream V4 editing at up to 4096x4096 — ByteDance's flagship, via API

One prompt, up to fifteen edited images — Seedream V4's batch mode

A whole Seedream series from one prompt — up to 15 images at once

The node every other node in this pack refuses to work without

ControlNet Union Pro 2.0 as a hosted call

Instruction editing with real guidance and step controls

FLUX.1 Kontext Max — the experimental tier, without the API paperwork

Kontext Pro, the API workhorse, as a plain ComfyUI edit node

Google's edit-anything image model, from inside ComfyUI

Nano Banana Pro editing, up to six reference images, no Google account

Nano Banana Pro edits that draw on up to six reference images

The 8K tier for when 4K isn't the goal, it's the floor

Nano Banana Pro T2I without the Google wrapper

Nano Banana Pro, two shots at once, zero GPU required

8K by API, and the node that prices it in dollars

The plain Nano Banana text-to-image node — the one you'll actually reach for

Veo 3.1 video from a still, with synced audio — the hosted shortcut

Google video, quickly, no GPU

Give Veo 3.1 a frame, get a movie — and an end-frame, if you want one

Keep the subject, swap the scene

The full-strength tier, not the fast one

The hosted AI upscaler that needs zero VRAM

A still photo that lip-syncs to your audio

Two People, One Photo, Two Audio Files — This Node Makes Them Talk

Animate a still image with Sora 2 for $0.40 — from inside ComfyUI

The paid, no-GPU route

Sora 2 from text alone — no image, no subscription, just a prompt and $0.40

Sora 2 Without an OpenAI Sub — Text-to-Video From Your ComfyUI Graph

The hosted instruction editor

Qwen-Image-Edit with style LoRAs — hosted, so you never touch the weights

Three reference images, one edit, exact output size

Multi-image, 3 LoRAs, and custom sizes

20B text rendering with zero local install

Qwen-Image T2I with style LoRAs — test the community library without downloading it

Video Upscaling With No VRAM — the Pack's Smallest, Most Honest Node

Drive one video's motion into another image — WAN 2.2 Animate, hosted

Wan 2.2 image-to-video without the 24GB VRAM tax

WAN 2.5 video with full controls — the non-fast hosted option

WAN 2.5 is API-only — this node is your fastest ComfyUI on-ramp

The API-only model, one POST away

Wan 2.5 Has No Weights — This Node Is the Only Way to Run It From ComfyUI

Wan 2.5 video you can't run locally (unless you pay for it)