Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

strandsagents+2.techcrunch.startupfortune.Barely two weeks after TypeSafe AI introduced Jev, OpenAI, Amazon Web Services and Cloudflare have each put out their own decision models. These systems don't write text. They pick an answer from a fixed list and attach a probability to it, and the burst of releases has quickly turned a niche idea into a crowded race.startupfortune+1
The rush shows how fast the industry has moved on what TypeSafe calls "System One" models. They handle the quick, routine judgment calls AI agents make all day, such as whether a task is urgent, whether to continue or stop, and whether to allow or block an action.digitaltoday+1
TypeSafe AI was founded by Diogo Almeida, a former OpenAI researcher. The company released Jev in mid-September and describes its output as "calibrated decisions" rather than text, TechCrunch reported. In a launch post, Almeida said the models "output decisions with probabilities and confidence instead of words". TypeSafe says Jev matches existing LLMs on these tasks while running two orders of magnitude faster. It charges $0.042 per million tokens. Startup Fortune reported that the company came out of stealth with $40 million in seed funding led by DCVC and is drawing investor interest at a $10 billion valuation.typesafe+4
At its developer event in late September, OpenAI unveiled a limited release of a "Decisions API," built on its Luna model. "If you narrow the range of choices, you can achieve very fast speed while still maintaining image understanding, broad language support, and safety guardrails," CEO Sam Altman said.digitaltoday
Amazon's Amazon.com, Inc. Strands Agents team released Strands Decider 2B, an open-source model under the Apache 2.0 license. It runs on a local CPU or GPU and answers in tens of milliseconds. Built on Qwen3.5-2B, it returns choices, yes/no probabilities or scores in a single forward pass. AWS said developers already use small decision models for tasks such as model routing, tool selection, guardrails and policy classification.ground+2
Cloudflare released two open-weight models, Clef and Clef-flash, hosted on its Workers AI platform, along with a new reinforcement-learning fine-tuning service. The company says Clef beats Jev on accuracy and latency across dozens of benchmarks. It also reports a median latency of about 39 milliseconds for Clef-flash. Clef costs $0.24 per million tokens on Workers AI, roughly six times Jev's price, though users can download the weights for free.cloudflare+4
The open releases put pressure on TypeSafe's business model. Startup Fortune called the trend "commoditization by subsidy," arguing that the startup is selling API access while "two of the biggest infrastructure companies in the world released equivalent capability under a license that costs nothing".startupfortune
TypeSafe has not disclosed Jev's architecture.digitaltoday