Introducing Clef: our open-source decision models, and new RL fine-tuning platform
Cloudflare open-sources Clef and Clef-flash decision models under Apache 2.0, with vision input, 64k context, and a new RL fine-tuning platform.
Cloudflare released two open-source decision models, Clef and Clef-flash, hosted on Workers AI and published on Hugging Face under Apache 2.0. Clef adds a vision encoder and a 64k context window versus Jev's 32k text-only design, and leads the Jev Decision Index benchmarks, with Clef-flash posting 38.8ms median latency versus Jev's 524.1ms. In an internal threat-intel workflow, Clef classified a website domain in 2.2s versus 4.7s for gpt-oss-120b. A new reinforcement learning product lets customers fine-tune Clef for their own use cases.
- Clef and Clef-flash open-sourced under Apache 2.0 and hosted on Workers AI
- Vision encoder and 64k context exceed Jev's 32k text-only limits
- Classified domains in 2.2s versus 4.7s for gpt-oss-120b in threat intel tests
- New RL fine-tuning product allows customer adaptation
- Tops Jev Decision Index; Clef-flash median latency 38.8ms
- Organizations
- CloudflareTypesafe AIClefClef-flash
From the original article
10 min read at blog.cloudflare.com
We are introducing Clef and Clef-flash, open-source decision models hosted on Workers AI for high-speed classification and agentic workflows. Also launching: a new reinforcement learning platform that allows developers to fine-tune decision models using their own data.
By Michelle Chen · blog.cloudflare.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.