Decision model Jev draws attention, with near-zero claimed bulk classification cost
Host Claire Vo's self-test put 200,000 classifications at about $4; the pricing model is worth watching for classification-heavy workflows.
ImportanceLocalEvidenceE2 unreplicatedWrite-upQuick
Per the show notes of the September 28 episode of How I AI on Lenny's Podcast Network, host Claire Vo tested Jev, a new decision model from TypeSafe AI, and reported processing 200,000 classifications for about $4.
Jev does not generate text; it returns predefined categories, scores, or probabilities. The notes say it costs 4 cents per million input tokens with no output-token fee. She reported spending 9 cents to analyze 1,700 pull requests, and also classified 4,500 YouTube comments.
These figures are a host's self-test as relayed in the episode notes; the pricing and product details have no vendor announcement or independent source yet, so readers running classification or routing workflows should watch for confirmation.