jevWhat this site is
An independent educational resource about TypeSafe Jev. It is not an official documentation mirror, a provider dashboard or a benchmark lab. TypeSafe AI and other names belong to their respective owners. This site is not affiliated with or endorsed by TypeSafe AI.
How we write
Technical claims link to official documentation wherever possible. Community integrations are identified as community maintained. Guides were reviewed against available sources on 20 September 2026. Always check upstream changes before shipping.
What we have and have not measured
The Request Builder generates local request files and code. The Evaluation Bench computes metrics from the predictions you supply. Neither calls Jev. Synthetic demo rows illustrate the workflow; they are not model results. No live API or head-to-head GPT/Claude benchmark has been run for this site. Our local lab release exercised 27 synthetic cases and 18 response-validation and injected-failure tests. Those tests verify application logic, not provider behavior.
Your data stays in this page
The tools process inputs in browser memory. This application has no analytics, account database or API-key collection. Reloading clears your work. Download files to keep them. External documentation links and the hosting platform have their own policies.
Scope of this first edition
24 guides, two local tools and five downloadable Python labs. The five task guides cover local application logic, failure handling and evaluation. Integration notes are labeled where a deployed provider setup has not been tested. The Videos page plays a community demonstration and links to official videos that restrict embedding, with creator credits and viewing notes. These external demos are separate from our locally tested labs. We do not host or re-upload creators’ videos.