#typesafe
Chen said Clef took about two weeks to build, while TypeSafe AI spent two years on Jev. She credited TypeSafe for inventing the approach.
November 18, 2026 at 12:23 PM
What happens when you constrain an LLM to structured choices and millisecond latency instead of open-ended text? Decision models are one answer.
AI Companies Develop Decision Models to Avoid Hallucinations
TypeSafe AI’s model Jev spearheads an effort to make cheaper, faster, reliable models
spectrum.ieee.org
November 16, 2026 at 5:02 PM
Decision models are just getting started.

As you might’ve heard, a start-up called Typesafe.ai released a new type of AI model, called Jev. They call it a System One model, while others have taken to calling them decision models.

spectrum.ieee.org/typesafe-ai-...
AI Companies Develop Decision Models to Avoid Hallucinations
TypeSafe AI’s model Jev spearheads an effort to make cheaper, faster, reliable models
spectrum.ieee.org
November 16, 2026 at 6:15 PM
TypeSafe AI's Jev sparks wave of fast, cheap decision models from Cloudflare, Amazon, OpenAI and Microsoft that cannot hallucinate
AI Companies Develop Decision Models to Avoid Hallucinations
TypeSafe AI’s model Jev spearheads an effort to make cheaper, faster, reliable models
spectrum.ieee.org
November 17, 2026 at 2:24 AM
Jev(TypeSafe)の Choice・Score・Noul・Box を試した所感 | m. k. #zenn
https://zenn.dev/mk404/articles/2026-11-17-jev-choice-score-noul-box-notes
Jev(TypeSafe)の Choice・Score・Noul・Box を試した所感
zenn.dev
November 17, 2026 at 2:57 PM
How much nuance there is in the API (more is worse, agents perform better with fewer choices)
5b. How typesafe it is
5c. How easy-to-test it is
5d. How much cyclomatic complexity it introduces
5e. The quality of docs provided
5f. What changes to deployment will be needed
6. (3/4)
November 17, 2026 at 9:11 AM
This post by @jola.dev was cool to read! Making scopes typesafe feels inches away from a form of RBAC (effectively)
Proposal for Tranquil PDS: Type-safe scopes
Tranquil is a popular AT protocol PDS implementation. This is a proposal to introduce a type-safe building block for representing scopes.
jola.dev
November 18, 2026 at 12:20 AM
How Decision Models Make AI Hallucination Vanish – A startup called TypeSafe AI released a new AI model called Jev. The company claimed that Jev, which it calls a “System One model” is cheaper, faster, and can’t hallucinate. Cloudflare, Amazon, OpenAI... https://tinyurl.com/23ctt3d2 #AIResearchers
AI Companies Develop Decision Models to Avoid Hallucinations
TypeSafe AI’s model Jev spearheads an effort to make cheaper, faster, reliable models
spectrum.ieee.org
November 16, 2026 at 6:20 PM
Cloudflare is launching two System One models, Clef and Clairic. Clef is a decision model that Chen says was inspired by TypeSafe AI's Jev model, which she described as an east wind that changed how Cloudflare viewed AI models.
November 18, 2026 at 12:21 PM
OpenAI is pricing AI decisions at $0.10 per million tokens and claims up to 10× lower latency than its Responses API. The bet: software will call models to choose, not explain. #OpenAI https://spaisee.com/article/openai-and-typesafe-turn-ai-choices-into-a-new-race-for-speed
OpenAI and TypeSafe Turn AI Choices Into a New Race for Speed
OpenAI and TypeSafe are competing to make AI-powered decisions faster, cheaper and easier to automate, turning confidence-scored choices into a new software infrastructure battleground for developers.
spaisee.com
November 15, 2026 at 7:03 AM
is zig metaprogramming typesafe, greatest forum thread locked after thousands of pages, etc
November 11, 2026 at 7:44 AM
AI Companies Develop Decision Models to Avoid Hallucinations - IEEE Spectrum apple.news/AvXecq5GdS1W...
AI Companies Develop Decision Models to Avoid Hallucinations — IEEE Spectrum
TypeSafe AI’s model Jev spearheads an effort to make cheaper, faster, reliable models
apple.news
November 17, 2026 at 2:26 AM
TypeSafe models handled over 1.5 billion queries last week as decision models multiply. Microsoft rolled out Decision-1.1 on Azure, while two-person startup Sqwish raised $3 million for its own model.

#AI
Decision models go mainstream, from Azure to a two-person startup
AI & Models · technopolis.news
technopolis.news
November 18, 2026 at 2:07 PM
Gemini Decision 1 outputs class probs + confidence, not tokens. $0.10/M input, native video/PDF, 256k ctx, 1k Qs/call. Beats TypeSafe Jev on mod edges. Replaces gen w/ single forward pass for agents. https://blog.google/innovation-and-ai/technology/developers-tools/gemini-decisions-api/
November 14, 2026 at 11:12 AM
it always has and will be easier to get software written in a typesafe language if you want a safe program.
November 11, 2026 at 6:34 PM
igual me refiero al "concepto" detras de jev y no a lo que fue jev en realidad

que tambien fue muy hypeado por demas

y que tampoco es la solucion a todo, tambien puede ser envenenado
TypeSafe Jev Fact-Check: The 200x AI Hype vs What's Proven
YouTube video by Prism Labs
youtu.be
November 18, 2026 at 6:24 PM
Everything else in this release, including Harbor experiment metadata and faster retention sweeps:
arize.com/docs/phoeni...

pip install -U arize-phoenix arize-phoenix-evals
11.18.2026: Python Decision-Model Evaluators and OpenTelemetry JS 3 - Phoenix
Python classification evaluators run on OpenAI and TypeSafe decision models, @arizeai/phoenix-otel moves to OpenTelemetry JS 3, and px lists prompt and dataset versions and tags prompts.
arize.com
November 19, 2026 at 12:57 AM
A TypeSafe atendeu mais de 1,5 bilhão de consultas na última semana com seus modelos Jev no OpenRouter. A Microsoft lançou o Decision-1.1 no Azure e a Sqwish captou US$ 3 milhões para disputar a infraestrutura de escolhas rápidas.

#IA
Modelos de decisão se tornam mainstream, do Azure a uma startup de duas pessoas
IA e modelos · technopolis.news
technopolis.news
November 18, 2026 at 11:07 AM
#^Rozhodovací modely slibují rychlejší AI bez halucinací, chyby ale nezmizely

Startup TypeSafe AI představil model Jev, který místo volného generování textu vybírá odpověď z předem určených možností. Díky jedinému průchodu modelem má reagovat výrazně rychleji […]

[Original post on libera.site]
November 16, 2026 at 5:11 PM
Google za osm týdnů zkopíroval rozhodovací API od Jev, včetně jeho vlastního slovíčka pro ano/ne otázku. Router napíšete jednou. Jenže jedna věc se při výměně modelu nepřenese, a právě na ní stojí vaše pravidla.
Google okopíroval rozhodovací API modelu Jev do posledního slova. Vaše prahy jistoty ale výměnu mode
Gemini Decision 1 je další rozhodovací API s pevnou sadou odpovědí, které přišlo během osmi týdnů od uvedení modelu Jev od startupu TypeSafe AI. Google od něj převzal i vymyšlený výraz pro otázku typu ano/ne. Z rozhraní se tak stává standard: kód napíšete jednou a dodavatele pak měníte podle potřeby
aifounders.cz
November 16, 2026 at 7:30 AM
system_one_client 0.4.0 [updated]

Client for System One decision models (TypeSafe Jev, Cloudflare Clef, OpenAI Decisions, Gemini Decisions, OpenRouter, self-hosted): typed Choice, Noul and ⋯

Licenses: Apache-2.0
November 14, 2026 at 1:00 AM
How Jev Decides About Images, With 4 Python Examples Note: This expanded article builds on the concept of Jev image decisions, drawing on TypeSafe.... @thecosmicmeta.com

https://thecosmicmeta.com/how-jev-decides-about-images-with-4-python-examples?utm_source=cosmicmeta&utm_medium=bluesky
How Jev Decides About Images, With 4 Python Examples
Note: This expanded article builds on the concept of Jev image decisions, drawing on TypeSafe and OpenRouter documentation, practical Python
thecosmicmeta.com
November 13, 2026 at 8:10 PM
TypeSafe AI
Fonts: Inter, JetBrains Mono, Die Grotesk C and LisaTerminal Paper 2X3Y

https://maxibestof.one/websites/93964-typesafe-ai
November 10, 2026 at 12:02 PM