Comfy Router is live One API for frontier image, video, 3D, and audio models. Same model string. Same arguments. No new SDK, no new key, no redeploy. What Comfy Router gives you: → Explicit routing. You name the provider, we call that provider. It's down? The request fails there. No silent fallback. → Every job returns the provider that ran it. Log it, bill it, debug it. → Async. submit() returns a request ID immediately. The queue retries 429s and transient errors until a slot opens. subscribe() submits and polls to completion. → Batch-friendly. Queue a few hundred jobs, hold the IDs, pull results as they land. Nothing blocking on a 5-min video render. → 24h retention on inputs and outputs, then deleted. → Comfy credits. No sub, no Router fee. Providers at launch: Comfy. Runware, Wavespeed, Fal, Higgsfield. Multi-provider where the model supports it. Get Your API Key with the link below. ⬇️

Hy Image3.5 preview is now available in ComfyUI. Professional-grade image generation, +30% win rate in human eval vs Hy Image3.0 → Text to image and Image to image in one model, up to 2K → Multilingual text, symbols, and small print that render correctly → Cinematic, comic, commercial photography, and illustration styles → Identity and product features that hold through scene, outfit, and style changes

Qwen-Image-2.1 is now supported in ComfyUI! Open weights. One 7B checkpoint that generates and edits. → Image generation at native 2K → Instruction editing from up to 10 reference images in a single pass → RGBA output, alpha included

FastH3 by FastVideo is now available in ComfyUI Video and native stereo audio, generated together, in seconds. Best for: → Previz and animatics that need lots of takes, quickly → Timing and dialogue tests before committing to a full-quality render → Short-form and social work on tight turnarounds Run it locally in ComfyUI and soon on Cloud ⬇️

In 9 days, OpenAI switches off the Sora 2 API. Sept 24: sora-2, sora-2-pro, every snapshot. Gone. If your workflow was prompt in, clip out, you're rebuilding. If the model was one node, you're swapping it. Models are rentals. Workflows are property.

The idea started while I was in Berlin last week. I shot the main video in my hotel room on my phone, and the rest came together using a mix of different tools. I’ll list them all below.

See the best @minimax_AI H3 workflows in action as we announce winners of our #ComfyH3# Sync Sound challenge on a special livestream with guest judges @peterom, @emmacatnip, @yachimat_manga! Tune in Thurs 9/3 at 10am PT

One week left in the ComfyUI x @MiniMax_AI Community Challenge. Four ways to win an RTX 5090 or RTX 5060 Ti. Fully virtual and open to everyone - local rig or Comfy Cloud. 9/1 — submissions close @ 9pm PST 9/2 — livestream with guest judges and winner announcements Share on your socials (X, Instagram, LinkedIn, YouTube, TikTok) tagging @ComfyUI @MiniMax_AI #ComfyH3# so we can reshare our favorites! Submit your video below 👇

84% of marketers use AI in their video work. The entire AI video generation market is ~$900M. The gap is the whole story. 90% of those marketers say the non-negotiable is editing what the model gives back. Everyone priced inference. Nobody priced that.

Wan 3.0 is now live on Comfy! It generates what earlier versions couldn't: a full 30-second scene in one take, as low as 480p for speed and iteration, and up to 1080p for production quality. Click the link below to try it today 👇

Livestream alert! Local MCP: Run ComfyUI with your Agent & Hardware on Weds. 8/26 at 10am PT. Join us for a live conversation on how to get the MCP running on your own machine, debugging workflows, and optimizing @MiniMax_AI H3 for your specific hardware.

We're teaming up with @MiniMax_AI on a community challenge built around H3. Two weeks, four ways to win an RTX 5090 or 5060 Ti, local rig or Comfy Cloud. 8/20 — challenge opens 8/20–9/1 — build and submit 9/2 — livestream with guest judges and winners Share on your socials (X, Instagram, LinkedIn, YouTube, TikTok) tagging @ComfyUI @MiniMax_AI #ComfyH3# so we can reshare our favorites! Click the link below to learn more 👇

Comfy MCP is now local and open-source! The #1# ask after Comfy Cloud MCP shipped in June. Your agent sees your actual install. So every node, custom nodes included, and every model on disk. It fetches files, starts your instance, gets a workflow to the point where it runs. It is now the easiest way to help with your local Minimax H3 workflows! Cloud MCP still does everything it did. Tell your agent where a job goes, or let it decide.

Coca-Cola and OpenAI both raised their US billboard spend by $2M+ last quarter. So did Pepsi, Uber, eBay, and Capital One. $2.12B quarter, twentieth straight of growth. The step that hasn't kept pace is showing a client what actually goes up. Every campaign on that list had to be comped first: artwork mocked onto a highway sign, a shelter panel, the side of a building, so someone could see it in place before committing budget. When that step is expensive, only the concepts a team is already confident in get shown. The rest aren't rejected. They're never presented. We published a workflow this week that takes one poster and returns three placements in a single run, each with its own scene prompt, under one shared control for time of day and lighting. The parallelism is the part worth noting. Holding one variable constant across three branches while the others differ isn't something a model gives you. It's a property of the pipeline.