DigestAI news desk

Cut through the AI noise.

Listen
Generative AI & Modelsupdated 7 min read

TypeSafe AI launches Jev, a decision model claimed up to 193.6x faster and 444.6x cheaper

TypeSafe AI announced the release of Jev, its first “System One” model that returns typed probabilistic decisions instead of free‑form text. The company says the model avoids hallucinations and is built with a training method called Reinforcement Learning for Calibrated Decisions (RLCD). All four sources note that Jev is positioned for automation tasks where speed and cost matter.

49 sources primary source HN 114

What you can do with it

AI at Work →
Jev API by TypeSafe AI

Route tickets to departments quickly with Jev API

Returns a chosen value, probability, and confidence score for set options.

Who for
founders and operations
Cost
paid from $42 per billion tokens
Effort
minutes

Use it for

  • Route tickets to departments
  • Triage emails automatically
  • Guard AI tool calls

Watch out Check the details on the maker's page first

Open Jev API ↗ What happened · Towards Data Science, finance.yahoo.com Run the business

Jev API in the tool directory

Key points from the news

  • TypeSafe AI released Jev, a “System One” model that returns typed probabilistic decisions rather than free‑form text.
  • Speed claims vary: 193.6x faster, 20–200x faster, >100x faster, and 70ms-500ms response times reported.
  • Pricing is said to be $0.042 / MTok input, $42 per billion tokens, or up to 444.6x cheaper than GPT‑5.6 Terra.

The reported performance and pricing numbers differ. The blog post cites "193.6x faster" and "444.6x cheaper" on its home page, while a separate analysis lists "20–200x faster" and "40–400x cheaper" and even " >100x faster, >200x cheaper". Another report gives concrete latency of "70ms-500ms" with a benchmark of "0.114s" versus "8.566 seconds" for OpenAI’s GPT-5.6 Terra, and mentions a "40x-200x faster" claim. Cost figures range from "$0.042 / MTok" for input versus "$2.00 / MTok" and "$12 / MTok" for output on GPT-5.6 Terra, to "$42 per billion input tokens" and a range of "$0.20 to $10 per million" for other models. Only The Register notes that TypeSafe AI has received "$40 million" in funding. The various multipliers ("193x faster", "444x cheaper") also appear in a separate piece, reflecting the same underlying claim with slightly different rounding.

The story so far

4 episodes →
  4. TypeSafe AI launches Jev, a decision model claimed up to 193.6x faster and 444.6x cheaperthis story
Full story from typesafe.ai · via Hacker NewsOpen source ↗

Jev: New frontier model 40-400x cheaper and 20-200x faster

typesafe.ai · 15 September 2026

Loading the full article…

This text was published by typesafe.ai. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Coverage and discussion

49sources
Topics · follow one to build your own front page
TypeSafe AIOpenAIOpenRouterJevGPT-5.6 TerraFable 5.1AstraDiogo Almeida

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.

Comments

via GitHub Discussions

More in Generative AI & Models

All →

Related stories

hot Proton CEO Andy Yen cancels AI re-interviews after employee backlash new Terence Tao warns AI is destroying mathematics' deliberative process hot OpenAI seeks $30 billion funding round at $1.25 trillion valuation new DeepSeek launches Super-Agent to automate developer tasks without human input hot Cisco Talos reports new VOLTA-AUTODOXER malware uses AI to rank victims by financial value new Microsoft and Cloud Essentials host webinar on AI adoption barriers hot Anthropic files for IPO, faces potential junk credit rating and $800B spending plan new OpenAI, Google, Microsoft propose common robots.txt rule to block AI training