Cloudflare launches Clef decision models to compete with TypeSafe’s Jev
Cloudflare released Clef and Clef-flash, two decision models designed to route AI agents’ actions without human oversight. Unlike traditional language models, these models assign probabilities to predefined options—such as ticket routing or escalation—enabling automated responses. Clef-flash responds in 39 milliseconds, while Clef takes 209 milliseconds, both faster than TypeSafe AI’s Jev, which…
Key points
- Cloudflare’s Clef models route AI agent actions in **39–209ms**, faster than TypeSafe’s Jev at **524ms**
- Both support text and images, with a **64K-token window**, double Jev’s capacity and outperform it on benchmarks
- Cloudflare offers fine-tuning via a reinforcement learning service, with self-service coming later
Cloudflare built Clef and Clef-flash using Qwen3.8-27B and Qwen3.5-9B models, respectively, and trained them with synthetic data and a variant of Reinforcement Learning for Calibrated Decisions (RLCD), the same method TypeSafe used for Jev. The company is rolling out a reinforcement learning service to let customers fine-tune the models for their own use cases, starting with engineer-assisted deployment before a self-service platform arrives. Clef is available on Hugging Face under Apache-2.0 and will be used internally for abuse reporting, support routing, and bot classification. Cloudflare’s move follows OpenAI’s Decisions API, also launched in late September, which similarly routes AI actions based on predefined options.
Model pages: Jev → · GPT-6 Luna →
The story so far
3 episodes →- Cloudflare launches Clef decision models to compete with TypeSafe’s Jevthis story
Cloudflare says its new Clef model means humans no longer need to be in the loop for AI agents
The Decoder · 2 October 2026
Loading the full article…
This text was published by The Decoder and written by Jonathan Kemper. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Gemini may delete existing text when asked to append, author reports · 1 src
- Google's Gemini outperforms ChatGPT and Claude as Android assistant · 1 src
- Meta launches Muse AI agent in Mexico, first in Latin America · 11 src
- Meta rolls out real-time voice and expressive avatars for Muse AI in the US · 7 src
- Meta improves Muse and Meta AI with Active Memory and Active Focus · 2 src
Comments
via GitHub Discussions