TypeSafe AI Ships Jev 1.14.0 with Image Input and 262k-Token Context
TypeSafe AI is adding capability while telling users to distrust the numbers that would sell it, which is the unusual position of a company whose product is judged against benchmarks it does not believe in.
Reporting from 1 source: GIGAZINE.
TypeSafe AI released Jev 1.14.0 on October 14, 2026, adding image input and a way to raise context length to 262k tokens per request. The decision model, which picks from preset options with a confidence score rather than generating text, had multimodal support disabled because non-text input performed worse. Long context stays off by default. The team also posted self-run comparisons showing higher intelligence, better cost efficiency, and a lower error rate than its previous model and other decision models.
Jev 1.14.0 keeps long context off by default even though the model can handle far more, because reliability drops as context grows. Users opt in per request through a header, up to 262k tokens for both the request and the question.
Image input arrives configured through a search-and-replace function. TypeSafe AI's own caveat is blunt: every model, Jev included, does worse on multimodal data than on text, and breaking questions into the simplest possible elements helps most.
The release also adds box, a preview type for locating an object in an image by its four sides, and an opt-in feature for contributors to surface model inconsistencies using queries.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.