Voices
TechCrunch gets three named engineers on the record about Jev, and two of them disagree on the cost
Tim Fernholz's September 18 piece on Diogo Almeida, the ex-OpenAI researcher behind RLHF and now TypeSafe's founder, quotes Vercel engineer Pranit Sharma seeing 5-18x speedups with better accuracy after swapping out an OpenAI model, while Bryo AI CTO Nikhil Mudholkar says Jev ran 10-20x more expensive than Gemini for his workflow automation despite better confidence scores. Armin Ronacher's caveat is operational: the probability outputs are only useful if you actually discard the low-confidence ones and act only on high-probability results. The disagreement matters because TypeSafe's headline pricing assumes a token profile that not every workload has.
Source
↳ Follow the thread