Cloudflare debuts open-weight multimodal decision models Clef and Clef-flash, claiming they are smarter and faster than Jev, based on Qwen3.8-27B and Qwen3.5-9B

The Clef models can process images and video, expanding decision model capabilities beyond text. Their Jev-compatible API means you can likely swap them into existing workflows without significant changes.

Cloudflare has released two new open-weight multimodal decision models, Clef and Clef-flash, designed to compete with TypeSafe's Jev model. Announced on Thursday, Clef models are based on specially post-trained versions of Qwen3.8-27B and Qwen3.5-9B. They perform a prefill-only pass during inference and can answer yes/no, multiple choice, and ranking questions. Cloudflare claims Clef is faster and more accurate than Jev, capable of processing images and video in addition to text, and supports a 64k context window. Clef models are available hosted on Cloudflare's Workers AI for faster edge processing, or downloadable from Hugging Face under an Apache-2.0 license. Clef-flash requires a GPU with 41 GB of VRAM, while Clef needs 85 GB. The API is Jev compatible, allowing for easy integration.

Cloudflare's entry into the multimodal decision model space with Clef and Clef-flash signals a growing trend towards more capable and versatile AI tools that can handle diverse data types beyond text. By leveraging open-weight models based on Qwen, Cloudflare aims to provide a competitive alternative to closed proprietary solutions, potentially driving down costs and increasing accessibility for developers. The focus on speed and accuracy, coupled with the ability to run locally on high-end hardware, suggests an effort to cater to both enterprise needs for performance and open-source enthusiasts seeking greater control.

The direct competition with Jev, including API compatibility, positions Clef as a potential drop-in replacement, aiming to disrupt the current market for structured decision-making AI. The higher token cost compared to Jev, however, indicates a trade-off for enhanced multimodal capabilities and potentially superior performance, making the decision to adopt Clef dependent on specific application requirements and budget considerations. Further independent benchmarking will be crucial to validate Cloudflare's performance claims against Jev and other emerging multimodal models.