Cloudflare has released Clef-omni, a multimodal decision model built on Qwen3-Omni-30B-A3B. Unlike generative models, it is designed to score inputs—text, images, audio, and video—in a single call, which makes it suitable for classification, routing, and other decision tasks. The model is distributed under the Apache-2.0 license, allowing broad commercial use.
The company also offers Clef-flash, a lower-cost variant. Clef-omni is priced at $0.15 per million input tokens, while Clef-flash drops to $0.038 per million tokens. The source is a single announcement, so there are no independent benchmarks or third-party comparisons to weigh; the reported capabilities and pricing come directly from Cloudflare's own material.
Because the source is a product release, it does not discuss limitations or trade-offs. The main takeaway is that Cloudflare is positioning Clef-omni as a cheap, unified scoring layer for multimodal data, potentially lowering the barrier for applications that need to evaluate many types of content in one pass.