Microsoft's new models reportedly up to 80x cheaper and 50x faster

ChrisUniverse · x · 2026-11-22

X user @ChrisUniverse reports that Microsoft (@MicrosoftAI) has released 7 new models, with some performing up to 80x cheaper and 50x faster (third-party claim, unverified). He argues that as model intelligence converges, differentiation will come down to accuracy, speed, and cost.

He also promotes Perplexity: it isn't constrained to a single model family, can run fully locally, and keeps private information away from 5-7 companies—@AskPerplexity can lock your data down locally or at least with a single company.

Original post →

More from Models

Models channel →