What is Clef?

Clef is Cloudflare's family of open-weight decision models, launched on Workers AI on 1 October 2026. There are two: Clef, a 27B model tuned for accuracy, and Clef-flash, a 9B model tuned for speed. Both do the same job — take application state plus a schema of typed questions and return a calibrated probability for each option in a single pass, no prose. They're Cloudflare's answer to Jev, the hosted model that opened the category. Here's what Clef is, how to run it, and where it sits next to Jev.

A decision model doesn't write text — it decides. You hand it some state (a ticket, a message, a row) plus a finite set of answers, and it returns which one, with a probability attached, in one forward pass. Clef is Cloudflare's open-weight take on that idea; Jev is TypeSafe AI's hosted, purpose-built one. Same call shape, different route to the answer.

The two Clef models

Cloudflare describes Clef as its first in-house open-weight decision models. Rather than generating free-form text, each returns strictly typed output — a probability for every allowed option of every question you send — in a single forward pass. Both are multimodal (text, JSON, and up to four images per request) and both ship as open weights on Hugging Face under Apache-2.0, so you can self-host them or call them on Workers AI as @cf/cloudflare/clef and @cf/cloudflare/clef-flash. Per Cloudflare's post, they freeze a Qwen backbone and add a vision encoder.

Question types are the familiar three: yes/no, multiple-choice, and rankings. Cloudflare also ships an RL fine-tuning service so you can post-train Clef on your own labelled decisions — the open-weight counterpart to Jev's pinned, hosted versions.

How you run Clef

• On Workers AI — call @cf/cloudflare/clef or @cf/cloudflare/clef-flash from a Worker binding or the REST API, no GPUs to manage (see Cloudflare's model docs for the exact request shape).
• Self-hosted — pull the weights from Hugging Face and run them yourself for data residency, air-gapped use, or custom hardware.
• Fine-tuned — use Cloudflare's RL service to specialise a Clef checkpoint on your own decisions.

Clef vs Jev, in one line

Clef is open-weight and self-managed (or edge-hosted on Workers); Jev is a hosted, RLCD-calibrated model with pinned versions and a free browser playground. The call shape is nearly identical, so the real choice is open-weight-and-self-managed vs hosted-and-calibrated. We break the head-to-head down on the Clef vs Jev page, and the Clef-flash page covers the smaller model specifically.

Try the hosted side in one call

The fastest way to feel the category is to make one real decision. Jev's browser playground does it with no signup, and a jv_live_ key points your own code at the hosted endpoint — the same state-plus-typed-questions shape Clef expects:

// A typed decision from the hosted model — the same shape an
// open-weight decision model like Clef expects.
const res = await fetch("https://jevtypesafeai.com/api/v1/decide", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    Authorization: `Bearer ${process.env.JEV_KEY}`, // jv_live_...
  },
  body: JSON.stringify({
    state: { ticket: ticket.body },
    questions: {
      queue: { type: "choice", instructions: "Route this ticket.",
               criteria: { billing: "payments", bug: "something broken", sales: "pricing" } },
      urgent: { type: "noul", instructions: "Is it urgent?" },
    },
  }),
});
const { answers } = await res.json();
route(answers.queue.choice, answers.urgent.noul > 0.8); // calibrated, sub-cent

• Cloudflare blog: Introducing Clef — Cloudflare's launch post for Clef / Clef-flash, with its own benchmark table (vendor-reported)
• Workers AI model: clef-flash — Cloudflare's model page — 9B decision model, 64k context, $0.09/M input

FAQ

See also: Clef vs Jev · Clef-flash specifically · What is a decision model · Jev on Cloudflare Workers · Playground