Post
CommunityJev Media & Talks
JEV, por qué ha revolucionado todo
Alex dc's Spanish-language explainer on Jev for the Agentes IA newsletter: what the TypeSafe decision model is, what it is not, and how to try it, opening with the headline claims of 193x faster and 444x cheaper.
- Format
- Post
- Reading time
- 3 min
- Published by
- Community
- Author
- Alex dc
- Last verified
- Oct 2026
Used inJev for Software & DevTools
Highlights
- Frames Jev as the model that does not write: pass it a text and a closed question, and it returns one of your options with a probability.
- Headline claims: 193x faster and 444x cheaper; the cited launch thread says 20-200x faster and 40-400x cheaper.
- Examples in the post: is this email urgent, is this lead qualified, is this comment positive, and where should this ticket go.
- Names Diogo Almeida, one of the co-creators of ChatGPT at OpenAI, behind TypeSafe AI, and cites 75,000 likes on the launch thread.
- Published September 23, 2026 on the Agentes IA Substack by Alex dc.
Watch out
Grounded on the captured page text, which is truncated after the first section, plus the digest's meta description; the later how-to-try-it steps and any pricing detail were not captured, so figures like 193x faster and 444x cheaper are reported as the post's framing, not verified benchmarks. The post is in Spanish and this entry summarizes the captured portion in English.
Questions about this entry
Short answers drawn from the linked project or page.
A Spanish-language explainer by Alex dc, published September 23, 2026 in the Agentes IA newsletter, on what Jev is, what it is not, and how to try it. The headline leads with 193x faster and 444x cheaper, and the post walks from the text generation Jev rejects to the decisions it returns.
It does not write. You give it a content, such as a customer email, a ticket or a lead, plus a closed question, and it returns one of the options you defined, with its probability, in under half a second. The post's example: is it a complaint, over 95 percent complaint, done.
Because with a model like ChatGPT or Claude you have to ask for the answer, read it, validate it and convert it into something your code understands, which is what makes those calls slow and expensive when you only wanted a yes or a no. Jev skips the text and hands back a decision your software can act on.
The headline says 193x faster and 444x cheaper; the launch thread it cites says 20-200x faster and 40-400x cheaper, and the post notes output tokens are free. Treat these as the post's framing of TypeSafe's claims, since the captured text does not include the benchmarks behind them.
Is this email urgent, is this lead qualified, is this comment positive, neutral or negative, and does this ticket go to billing, support or sales. These are the everyday routing and classification judgments the post contrasts with asking a chat model to write an answer.
TypeSafe AI, with Diogo Almeida, described in the post as one of the co-creators of ChatGPT at OpenAI. The post also cites 75,000 likes on the launch thread and says the model dominated AI news for seven days after release.
The piece is in Spanish on the Agentes IA Substack. This entry summarizes the captured introduction, which ends mid-section; the promised steps for trying it this week were not in the capture, so check the post itself for the walkthrough.
More Jev resources like this
Microsoft's announcement of Microsoft-Decision-1, a model for fast decision scoring built for routing, classification, prioritization, verification and workflow control, available in Microsoft Foundry and through OpenRouter. Published 2026.10.09.
Jev Media & TalksPost#community#catalog-import#madewithjev
Community
Cloudflare's announcement of Clef-omni, an open-weight decision model that takes audio and video alongside text and images in one request, built on a Qwen3-Omni mixture-of-experts backbone, plus a faster Clef and a cheaper Clef-flash.
Jev Media & TalksPost#community#catalog-import#madewithjev
Community
Anton Zagrebelny's analysis: Jev's free output tokens and confidence-per-decision design push products toward two-model architectures and quietly break the way AI credits and entitlements are metered.
Jev Media & TalksArticle#community#catalog-import#yibie
Community
A correctness-validated Laya runtime for Apple Silicon that runs upstream Laya on the MLX GPU and the Apple Neural Engine, using the ANE only after an artifact passes a parity gate on your machine, plus a local Jev-compatible server.
Jev Repos & SDKs#community#catalog-import#madewithjev
Community
tc3oliver/laya-applePython
tc3oliver/laya-applePythonLaya GGUF inference and an HTTP server written in Goish Rust: one Linux binary runs upstream Laya or OpenThai-SystemOne on your CPU and serves typed choose, score and check decisions over HTTP, the CLI or a JSON-lines process.
Jev Repos & SDKs#community#catalog-import#madewithjev
Community
cogentica-ai/laya-goishRust
cogentica-ai/laya-goishRustA tuner that trains language models to answer closed-set decisions with a complete probability distribution in one forward pass, using a tokenized multi-class Brier loss so only the correct choice is needed to train.
Jev Repos & SDKs#community#catalog-import#madewithjev
Community
liushiliushi/jevtunerPython
liushiliushi/jevtunerPythonFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.