ComfyUI, a startup that provides creators with granular control over image, video, and audio outputs from diffusion models through a node-based workflow, has secured $30 million in funding at a $500 million valuation. The investment round was spearheaded by Craft Ventures, with additional backing from Pace Capital, Chemistry, and TruArrow.

The company began as an open-source project in 2023, soon after diffusion models emerged. At the time, platforms like Midjourney and OpenAI’s DALL-E were still quite rudimentary, often producing obvious errors—such as generating extra fingers on hands. To tackle these shortcomings, the project's founders built a modular framework that enables creators to manage each step of the generation process with precision. The tool quickly gained popularity among creative professionals, eventually evolving into a formal startup. In late 2024, ComfyUI raised $19 million in Series A funding from investors including Chemistry Ventures, Cursor Capital, and Guillermo Rauch, the founder of Vercel.

While today's diffusion models have largely moved past blunders like adding a sixth digit to hands, the demand for the kind of detailed control ComfyUI offers has only intensified. "But to change that remaining 20%, you have to try this slot machine," said a company representative, comparing the process to gambling in a casino. This is because prompting a model for a small adjustment can lead to an entirely different result, often overwriting parts that were already flawless.

ComfyUI's node-based interface lets creators connect specific elements of the generation pipeline, giving them full authority over the quality of the final output. "You cannot easily convey that message in the prompt box [of a foundational model]," the representative added. Creators appear to agree, as ComfyUI reports having more than 4 million users. The tool is employed by professionals in fields such as visual effects, animation, advertising, and industrial design.

The startup asserts that its offering has become an essential tool for technical artists and other creatives, to the point where "ComfyUI artist or engineer" is frequently listed as a job title on studio hiring boards. Even as video and image foundational models continue to advance, the company argues they are still far from flawless, ensuring that a tool like ComfyUI will remain in high demand. "In the world where AI slop is going to be everywhere, the Comfy version of human-in-the-loop approach is going to win out most of the eyeballs in the end," the representative said. Competitors in this space include Weavy, a startup acquired by Figma last year.