Free AI Art Tool ComfyUI Now Runs GPT, Claude and ByteDance Models

ComfyUI is a free, open-source program that lets people generate AI images and videos on their own computer. Think of it as a workshop where you connect different AI tools together like building blocks. It has 136,000 stars on GitHub, a rough popularity score, making it one of the most-used AI art tools in the world. Version 0.38.0, released in late September, is its biggest update in a while.

The headline change is a batch of "partner nodes" — ready-made plug-in blocks for outside companies' AI models. You can now drop in Anthropic's Claude Opus 5.5 for writing help, OpenAI's GPT-6 Sol and Luna models, ByteDance's Seedream 5.0 for images and Seedance 2.5 for video, Tencent's Hunyuan Image 3.5, and Quiver's Arrow 2. Normally, connecting these services means wrestling with code and settings. Now it's a drag-and-drop step. The update also removed older Sora nodes that no longer work.

The quieter half of the release is about speed and memory. Dozens of small fixes make models load faster, use less of your computer's memory, and stop crashing on unusual files like transparent images. One change even lets the software notice whether a file is on a fast or slow drive and adjust accordingly. There's also new support for the Qwen Image 2.1 family, a recipe for blending video frames more smoothly, and better color handling for professional video work.

Why does any of this matter if you're not an artist? Because ComfyUI is the free alternative to paid AI subscriptions. When it adds support for paid models, you still pay those companies per use — but you keep the tool, your files, and your privacy. And the performance work means people with ordinary laptops, not expensive gaming rigs, can actually run modern AI generation at home.