Tech companies move to open AI models

The Pragmatic Engineer · 1 month ago · 49 ● 9 sources

Tech companies are shifting AI workloads toward open models to reduce inference costs while maintaining usable output quality. Uber reported cutting cost per AI request by 34% and cost per AI session by 52%, keeping March costs flat despite higher usage. This change drives more benchmarking, model routing, cheaper open-model deployment, and prompt/request optimizations to lower spending without a noticeable quality drop.