TypeSafe AI · 2026-10-14 · notable
Jev 1.14 — TypeSafe's decision model adds images and bounding boxes
Jev 1.14.0 is TypeSafe AI's new decision model, out now as jev-preview. It brings small gains across the board, mostly in multilingual support, plus opt-in experimental image input, a Box question type and a 262K long-context mode.
TypeSafe's typed-answer model gets better multilingual support and experimental images, boxes and long context.
What is it?
Jev 1.14.0 is the second release of TypeSafe AI's System One decision model, which answers typed questions instead of writing text. TypeSafe calls it "small, subtle improvements to everything", with the largest expected gain in multilingual support. It ships under the jev-preview alias, and jev-latest moves to 1.14.0 in one week.
How does it work?
The new abilities sit behind opt-in "danger zone" API headers. With TypeSafe-Danger-Zone-Images, PNG or JPEG images go in a request's assets map and are referenced from the state, instructions or criteria. A new Box question, only on Jev 1.14, returns one bounding box as four edges plus a confidence. TypeSafe-Danger-Zone-Long-Context raises both context limits from 64K and 32K to 262K tokens.
Why does it matter?
Developers who already use Jev for routing and triage can now ask it about screenshots or scanned forms, and point at where something is in an image. TypeSafe warns these features still have "way too much jaggedness" for production. Jev 1.13.0 stays available, with 60 days' warning promised before it is deprecated.
Who is it for?
developers using decision models in automation
Try it
Set model to jev-preview or jev-1.14.0 in the TypeSafe SDK or API