On July 9, 2026, OpenAI publicly released GPT-5.6, its most capable model family yet, in three tiers: Sol, the top model; Terra, a mid-range option; and Luna, a fast, low-cost tier. The company framed the lineup around efficiency as much as raw power — CEO Sam Altman said Sol is roughly 54% more token-efficient on coding tasks than the previous generation.
Agents and voice, not just chat
Alongside the models, OpenAI shipped ChatGPT Work, an agent built to carry out complete jobs rather than answer one question at a time, and GPT-Live, a full-duplex voice mode that can listen and speak at the same time for more natural conversations. Together they mark OpenAI's push from a chatbot toward systems that act on a user's behalf across enterprise work, coding and research.