Cloudflare says its new Clef model means humans no longer need to be in the loop for AI agents

Cloudflare has released Clef and Clef-flash, two decision models for AI agents that the company positions against TypeSafe AI's Jev. According to Cloudflare, a decision model returns a brief classification with probabilities rather than a long text response. Given a customer support message, Clef can assess its urgency and identify the team that should handle it, and downstream code can use those results to route a ticket, trigger an escalation, or hand the case to a human.

Cloudflare says a human does not necessarily need to be in the loop for agentic decisions anymore, and that agents can programmatically gather context, make decisions, take actions on tasks, or defer to a human when needed.

Across 43 benchmarks, Cloudflare reports that Clef and the smaller Clef-flash are faster than all relevant competing decision models. The company cites median latency of about 39 milliseconds for Clef-flash and about 209 milliseconds for Clef, compared with just over 524 milliseconds for Jev. Both models run directly on Cloudflare's infrastructure.

Clef is based on Qwen3.8-27B and Clef-flash on the smaller Qwen3.5-9B, according to Cloudflare. The company says it leaves the base models unchanged during training and uses its own synthetic data to train extra components that derive answer options and probabilities from the models' internal computations. Cloudflare also uses its own variant of Reinforcement Learning for Calibrated Decisions, the training method TypeSafe used for Jev.

Cloudflare says Clef can process images, while Jev is limited to text so far, and that its 64,000-token context window holds twice as much input as Jev's. Cloudflare's threat intelligence team is already testing Clef to classify websites; in one example it assigned a domain a 95 percent probability of being a fashion website, 85 percent of being an online store, and under one percent of being a phishing site. Fetching, rendering, and classifying the site took 2.2 seconds, compared with 4.7 seconds for Cloudflare's fastest general-purpose language model, which returned only two categories.

Cloudflare is also rolling out a reinforcement learning service for customers to tailor Clef to their own tasks, initially through forward deployed engineers with a self-service platform planned for later. Both models run on Workers AI and are available on Hugging Face under the Apache-2.0 license. Cloudflare says it plans to use Clef internally to review abuse reports, sort support requests, and distinguish useful bots from harmful ones.

TypeSafe AI, founded by former OpenAI researcher Diogo Almeida, introduced Jev in mid-September. In late September, OpenAI followed with a Decisions API built on GPT-6 Luna that also accepts context as text or images.