Cloudflare launches Clef multimodal decision model
Cloudflare 推出基于 Qwen 的开源多模态决策模型 Clef
Clef's integration of visual capabilities with extended context creates new possibilities for complex business decision systems. Developers can now build multimodal applications that process both text and images simultaneously, with Cloudflare offering customization options through their new reinforcement learning product. The model's performance on Jev benchmarks positions it as a strong alternative to existing decision-focused AI systems.
Cloudflare released Clef and Clef-flash, two open-source multimodal decision models based on Qwen.
Clef uses Qwen3.8-27B while Clef-flash uses Qwen3.
5-9B.In Jev Decision Index 0.
2.1 benchmarks, Clef leads competitors.
Both models feature a native visual encoder and 64k context window, double Jev's 32k capacity.
Cloudflare 推出基于 Qwen 的开源多模态决策模型 Clef
IT之家 10 月 2 日消息,当地时间 10 月 1 日,Cloudflare 推出基于 Qwen 的开源多模态决策模型 Clef(谱号),包含 Clef 与 Clef-flash 两款模型。 在 Jev 决策指数(Jev Decision Index)的基准评测中(Decision Index 0.2.1),Clef 处于领先地位;这两款模型完全兼容 Jev-API,方便开发者进行测试验证。 此外,Cloudflare 还发布了全新的强化学习产品。借助该产品,客户还可以针对自身的具体业务场景,对 Clef 进行定制化微调。 据悉,Clef 选用 Qwen 作为基座模型,其中 Clef 基于 Qwen3.8-27B、Clef-flash 基于 Qwen3.5-9B。 据介绍,Clef 原生配备了视觉编码器,能够直接解析图像输入并对视觉内容进行判别分类;而目前的 Jev 仅支持纯文本分类。Clef 将上下文窗口拓展至 64k(相比之下 Jev 为 32k),这使得开发者能够为模型载入更为丰富的系统输入状态,从而实现更全面的综合判定。 官方表示,Clef 具备高精度与强劲的综合性能,在多项基准测试中均展现出比肩甚至超越同类模型的竞争实力。基准测试详情参见下表: IT之家附开源链接: https://huggingface.co/Cloudflare/clef
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Clef, Qwen, Cloudflare, multimodal.
Source attribution: curated from IT之家.