OpenAI has unveiled GPT-5.6, a new family of AI models comprising three variants: Sol (its flagship "workhorse"), Terra (an intermediate option) and Luna (a budget-friendly tier). The launch intensifies competition in an increasingly crowded market, with the models pitched at enterprise work, coding and scientific research, and marketed heavily against rival Anthropic. OpenAI also introduced ChatGPT Work, an enterprise companion tool for desktop, web and mobile that assists with clerical tasks such as drafting documents, spreadsheets and presentations.

Chief executive Sam Altman said the models are far more efficient than their predecessors, claiming Sol is 54% more token-efficient on coding tasks. Citing the Artificial Analysis Coding Agent Index, OpenAI says Sol scores 80 — 2.8 points above Anthropic's Fable 5 — while using less than half the output tokens, running in under half the time and costing roughly a third less, and adds that Luna outperforms Opus 4.8. The company also calls 5.6 its strongest cybersecurity model yet, supporting defensive activities such as threat modelling, code review and blue teaming, having previously faced attempts by the Trump administration to restrict its rollout over misuse fears. GPT-5.6 is available across ChatGPT, Codex and the API, priced per million tokens at $5/$30 (Sol), $2.50/$15 (Terra) and $1/$6 (Luna) for input/output.