Jev

The model that reached a third of Vercel's paying teams fastest cannot write a sentence

Jev picks between a small set of predetermined outputs rather than generating text. Vercel's figures cover its own gateway, not the labs' direct APIs.

Cloudflare's decision model now reads audio and video, and the cheap tier fell to $0.038 with a smaller window

The new model is built on Alibaba's Qwen3-Omni with the speech output stripped out. Cloudflare says 0.24% of requests exceed the shorter window.

Microsoft's first decision model is a post-trained Qwen, costs $0.042 a million input tokens and charges nothing for output

The model is built on Alibaba's open Qwen3.5-9B, and Microsoft says it will later rebase the same system on its own MAI models and on OpenAI's.

TypeSafe AI has raised $870m at a $7.5bn valuation, three and a half weeks after shipping a model that does not write anything

The company calls Jev a System One model. It is still in early access, and the only usage figures TypeSafe has given are two sentences without numbers in them.

Reading a model's internals instead of its output cut the cost of watching for cyber misuse by about thirty times, Goodfire says

A lightweight probe reads activations at every turn and escalates only what looks wrong, which drops judge cost from $3,166 to $65 per million turns.

Perplexity has opened a 27-billion-parameter model that answers with a probability instead of a sentence

It is the third decision model released in two days, after Amazon's Strands Decider and Cloudflare's Clef. The category did not exist in March.

Stanford and Nvidia have opened a decision-only model that answers by matching, not by writing

The weights are on Hugging Face under Apache 2.0, and the trained part is two small projection heads on a frozen Qwen3-8B encoder.

A ChatGPT co-creator has shipped a model that writes nothing and only returns decisions

TypeSafe AI opened early access to Jev on 15 September after two years in stealth. The company says it matches language models on decision tasks while running two orders of magnitude faster.