Article
OfficialJev Guides & Articles
Jev: For the Builders AND Dreamers (Jev 1.14)
The Jev Day release post: Jev 1.14.0 under jev-preview plus three danger-zone features — composable images on both models, the box question type on 1.14 only, and opt-in 262k context.
- Format
- Article
- Reading time
- 3 min
- Published by
- TypeSafe AI
- Last verified
- Oct 2026
Highlights
- Jev 1.14.0 ships as jev-preview: small, subtle improvements everywhere, with the largest expected gain in multilingual support.
- Images are composable on jev-1.13.0 and jev-1.14.0 via find-and-replace in assets; the post confirms 1.13 was multimodal all along.
- box is the first type not meant for smart-if statements, and TypeSafe enables it only on jev-1.14.0.
- Danger-zone flags gate the experimental features; TypeSafe says there is still too much jaggedness to vouch for production use.
- jev-latest moves to 1.14.0 in a week; deprecation intent is 60 days' warning for 1.13.0 and 30 for 1.14.0, with a policy to come first.
Watch out
TypeSafe's own framing: the danger-zone features will stay on future models and improve, but the post warns they are not production-grade and that all models, Jev included, perform worse on images than on text. The internal intelligence measurement is self-described as likely biased toward TypeSafe's models, with no numeric table.
Questions about this entry
Short answers drawn from the linked project or page.
Jev 1.14.0 under the jev-preview alias plus three experimental danger-zone features: images on both current models, the box question type on 1.14 only, and opt-in longer context. The post frames them as dreamer-ready rather than production-ready.
The post explains the balance between reliability and creativity: TypeSafe does not want to impose its reliability bar on builders experimenting, so dreamer-ready features ship behind API headers. The features will stay and improve, but TypeSafe says there is too much jaggedness to vouch for them in production.
Yes, per the post: images are enabled for the current Jev that builders already know, and jev-1.13.0 was multimodal all along. Image support is composable via find-and-replace in the assets map, usable in state, instructions, or criteria.
It is the first TypeSafe type not purely meant for smart-if statements, returning a single bounding box as a preview of explorations beyond simple decisions. The post says TypeSafe made recent leaps on it and enables box only for jev-1.14.0.
Jev always supported a much longer context than production enabled; TypeSafe capped it intentionally because models are less reliable on long contexts and the constraint encourages better usage patterns. The post still strongly recommends limiting context length even though the 262k mode exists.
jev-latest updates to jev-1.14.0 about a week after the release; jev-1.13.0 stays available for a while. TypeSafe intends 60 days' warning before 1.13.0 is formally deprecated and 30 days for 1.14.0, and says a deprecation policy lands before either.
TypeSafe shares its own best guess at measuring developer-relevant intelligence, includes only open-weight models because it guards its validation set, and says the measurement will evolve, is likely biased toward its models, and should be ignored in favor of your own evals.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Jev Day teaser: what's this, Jev is evolving
X: Happy JevDay: images, box, longer context, and an opt-in to find jaggedness
X: Jev 1.14 is now on OpenRouter
X: Jev Day recap: four times the context and the danger zone
More Jev resources like this
The enterprise page for Jev: two published case studies with 88% and 98.7% lower cost, SOC 2 Type II and zero data retention, a no-training-on-inputs commitment, and higher rate limits for teams.
Jev Guides & ArticlesDocs#official#enterprise#compliance
Official
A Chinese-language summary of the Latent Space interview with TypeSafe CEO Diogo Almeida, framing Jev as a production System One model rather than a general God model: parallel sampling, 70–500 ms latency, and $0.042 per million input tokens.
Jev Guides & ArticlesArticle#community#article#interview
Community
Crunchbase profile for TypeSafe AI: CB Rank 25, San Francisco headquarters, a Series A in Funding Rounds, founders Diogo Almeida, Erik Gafni, and Sasha Sheng, and Jev described as a System One Model returning typed decisions.
Jev Guides & Articles#community#company#funding
Community
The companion source code for the book Mastering TypeSafe Jev: 13 chapters plus two appendices build two capstones on one shared decision client — an autonomous incident and FinTech triage engine on hosted Jev, and a SmartPlant industrial-IoT engine on self-hosted open-source Kev.
Jev Guides & Articles#community#catalog-import#yibie
Guidance on how Jev fits coding agents: it is not a drop-in agent model, so the page routes readers to routing, rubric scoring, and truth checks instead.
Jev Guides & ArticlesDocs#official#docs#coding-agent
Official
The idea behind Jev: models that return typed decisions and probabilities instead of generated text, named after Kahneman's fast-thinking System 1, with confidence to route around uncertainty.
Jev Guides & ArticlesDocs#official#docs#architecture
Official
From 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.