Introducing Clef: our open-source decision models, and new RL fine-tuning platform
Cloudflare has released Clef and Clef-flash, two open-source decision models integrated into the Workers AI platform. Additionally, the company launched a reinforcement learning (RL) platform for developers to fine-tune these models using proprietary datasets.
Verified State Diff
Impact & Verification Analysis
Developers building agentic workflows, AI engineers, and enterprise users leveraging Cloudflare Workers AI.
This release lowers the barrier for deploying high-performance, domain-specific AI agents at the edge, moving beyond simple text generation to actionable decision-making while keeping data within the Cloudflare ecosystem.
Full Fact Overview
The introduction of Clef and Clef-flash represents a strategic shift for Cloudflare from general-purpose generative AI toward specialized, high-speed classification and agentic decision-making workflows. By hosting these models directly on Workers AI, Cloudflare reduces latency for inference tasks. The inclusion of an RL fine-tuning platform allows for iterative model optimization, enabling developers to align model outputs with specific business logic or domain-specific data without needing external infrastructure.