Clef

Overview

Cloudflare released Clef on October 1, 2026, an open-source decision model post-trained on a frozen Qwen3.8-27B backbone with rank-256 low-rank adapters, adding a vision encoder and a 64k-token context window (versus 32k on Jev). Rather than generating prose, Clef takes a state plus a set of typed questions and returns calibrated probabilities over predefined answers, for tasks like routing, classification, and scoring inside agent workflows. Cloudflare reports a median latency of about 209ms across 43 internal benchmarks (versus 524ms for Jev) and says Clef outperforms Jev on 3 of 4 workflow evaluations while also reading images and video frames, which Jev cannot. Weights are open-sourced on Hugging Face under Apache 2.0 and hosted on Workers AI.

This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.

Key features

• Post-trained on frozen Qwen3.8-27B backbone with rank-256 low-rank adapters
• Returns strictly typed, non-autoregressive outputs with calibrated confidence scores
• 64k-token context window, double Jev's 32k
• Vision encoder enables image and video-frame classification, unlike text-only Jev
• Median latency ~209ms across 43 benchmarks, reported 2x latency savings over Jev in internal threat-intel testing
• Open-sourced on Hugging Face under Apache 2.0; hosted on Cloudflare Workers AI

Use cases

• Routing and classification steps inside agentic workflows
• Image/video-aware content and threat-intelligence classification
• Replacing LLM calls for bounded decision tasks to cut latency and cost

Pricing

Open-weight model, free to download under Apache 2.0; Workers AI hosted usage pricing not published in launch coverage.

Pricing reflects the source record and may change. Confirm current terms with the provider.

Dates and sources

Released

1 Oct 2026

Record updated

4 Oct 2026

Record last checked

4 Oct 2026

• blog.cloudflare.com/clef-decision-models/
• the-decoder.com/cloudflare-says-its-new-clef-model-means-humans-no-longer-need-to-be-in-the-loop-for-ai-agents/
• ai-tldr.dev/models/clef/