Clef Flash

Clef Flash is the small, fast version of Cloudflare's decision model Clef: instead of writing text, it returns a yes/no answer, a choice, or a score with a probability. It launched on the same day as the larger Clef and gives up some accuracy in exchange for speed and lower cost. Because it is built on a smaller Qwen model, it is light enough for services that make a very large number of decisions, and Cloudflare says it judges well for its size. It understands text, images, and video together, reads long documents in one go, and developers can plug it into their own products through Cloudflare Workers AI. It uses the same request format as Jev, so it is easy to swap with other decision models, and it was released so that anyone can download and run it themselves. On harder judgments it can be less accurate than Clef, so for work where accuracy matters it is worth testing both models side by side.