Builds Sage, a decision and extraction model API that lets software make calls at runtime
Levanto Labs builds Sage, which it describes as the first decision model: an API that lets a running program ask natural-language questions about text (and images in beta) and get back structured answers with calibrated confidence in around 100 ms. Its Sage2 release, launched on 2 November 2026, adds structured extraction and streaming decisions that update as new text arrives, for use cases such as customer support, sales calls, trading and games. Sage is also available on NEAR AI for confidential inference.