OpenAI just launched GPT-5.6, a new family of models with three tiers: Sol (premium workhorse), Terra (mid-range), and Luna (budget-friendly). They claim major leaps in efficiency—Sol is 54% more token-efficient for coding, and the whole family is touted as their strongest cybersecurity lineup yet, supporting defensive tasks like threat modeling and blue teaming. The Trump administration previously tried to restrict rollout over misuse fears. OpenAI also debuted ChatGPT Work, a workplace assistant for enterprise teams handling docs, spreadsheets, and presentations. The release directly challenges Anthropic, with OpenAI using the Coding Agent Index to argue Sol beats Anthropic’s Fable 5 by 2.8 points while using half the tokens, half the time, and costing a third less. Pricing runs from $5/$30 per million tokens for Sol down to $1/$6 for Luna. Why it matters: This aggressive pricing and tiered strategy signals OpenAI’s intent to dominate both enterprise and budget markets simultaneously, potentially reshaping the AI competition landscape. The model is available now across ChatGPT, Codex, and the API.