The efficient frontier of LLM inference Published 2026-09-02 · news The efficient frontier of LLM inference