Clef vs Jev

Clef and Jev are the two ends of the decision-model market in late 2026: Cloudflare's open-weight Clef (27B) and Clef-flash (9B) that you self-host or run on Workers AI, versus TypeSafe's hosted, RLCD-calibrated Jev that you reach through a managed API. Both take state plus typed questions and return calibrated probabilities in one pass — so this is an ops-and-calibration decision, not a 'can it decide' one. Here's the honest head-to-head.

If you're choosing between Clef and Jev, you've already decided you want a decision model rather than an LLM you prompt and parse. What's left is the trade-off: open weights you manage (Clef) versus a hosted, calibrated endpoint you call (Jev). The decide call looks almost the same either way.

Specs, side by side

The Clef columns use Cloudflare's published Workers AI figures; the Jev column uses TypeSafe's. Every number here is vendor-reported, not an independent rerun.

The benchmark claims — read them honestly

Cloudflare's launch post benchmarks Clef against Jev and reports wins on several classification tasks: BANKING77 macro-F1 94.20 vs 79.74, CLINC150+OOS 97.43 vs 89.27, BFCL case-exact 98.47 vs 95.75, and a median latency of 209.3ms against Jev's 524.1ms. It also says Clef beat Jev in three of four areas on TypeSafe's own workflow evals.

But the same post shows Jev ahead on others — When2Call 80.97 vs 72.37, BRIGHT 47.52 vs 45.91, agent-trace observability 71.6 vs 68.5 — so even by Cloudflare's numbers it's task-dependent, not a sweep. TypeSafe separately reports Jev at roughly 70–500ms per call, below Cloudflare's 524.1ms figure, which is a good reminder that latency depends on who's measuring. The only benchmark that settles your case is your decision, on your data.

When to pick which

• Pick Clef or Clef-flash when you want open weights — to self-host for data residency or air-gapped runs, to fine-tune on your own decisions with Cloudflare's RL service, or because your stack already lives on Workers and an edge call is the natural fit. Clef-flash if latency rules; Clef if accuracy does.
• Pick Jev when you want a hosted, managed model with RLCD-calibrated confidence and pinned versions, no GPUs or inference ops to run, image-free text/JSON decisions, and a drop-in API plus a free browser playground to feel the behaviour before you integrate.
• Not sure? Prototype on one and keep the other as a fallback — the call shape barely changes between them.

Migrating between them

Because both speak the same decide shape — state plus a schema of typed questions in, calibrated probabilities out — moving between Clef and Jev is mostly a transport change. Going Clef → Jev, you drop the model hosting and point at the hosted endpoint below, gaining RLCD calibration and pinned versions; going Jev → Clef, you take on inference ops in exchange for open weights and edge locality. Re-tune any confidence thresholds after a switch: calibration differs between a from-scratch RLCD model and an RL-fine-tuned open backbone, so an 0.8 cut-off may not mean the same thing on both.

Make one real decision on Jev

The fastest way to judge is to run your own decision through it. Jev's playground does it in the browser with no signup, and a jv_live_ key points your code at the hosted endpoint:

// Hosted, calibrated decision — the same state+questions shape 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 with its Clef-vs-Jev 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: What is Clef · Clef-flash specifically · Jev alternatives · What is a decision model · Playground