OpenAI today begins the global rollout of its GPT-5.6 model family — Sol, Terra, and Luna — marking one of the most anticipated AI releases of 2026. The three-tier system goes live for ChatGPT users and API developers starting July 9, with access expanding gradually.
Three Models, Three Purposes
GPT-5.6 Sol is the flagship: built for the hardest problems in research, software engineering, and multi-step reasoning. GPT-5.6 Terra targets everyday professional work — writing, analysis, brainstorming, coding — delivering GPT-5.5-class performance at roughly half the cost. GPT-5.6 Luna is the fastest and most economical tier, optimized for quick conversations, summaries, and routine tasks.
Deeper Reasoning, Multi-Agent 'Ultra' Mode
A key architectural change is that GPT-5.6 can spend significantly longer "thinking" through difficult problems before responding. OpenAI has also introduced an "Ultra" mode where multiple AI agents collaborate behind the scenes on complex requests — effectively parallelizing subtasks before synthesizing results.
Cybersecurity and Safety Upgrades
GPT-5.6 comes with substantially improved vulnerability detection and code review capabilities for defensive security work, paired with strengthened guardrails to prevent misuse. OpenAI says it spent weeks stress-testing the models using both human experts and automated red-teaming, adding a new "pause and review" mechanism for risky requests.
The Competitive Landscape
The launch arrives as Anthropic's Sonnet 5 (released June 30) and Fable 5 have raised the bar on agentic coding, while Google's Gemini 3.5 Pro remains delayed in preview. GPT-5.6's three-tier pricing — especially Luna's low-cost tier — is a direct response to the market pressure from Chinese open-weight models like GLM-5.2, which have captured up to 46% of US enterprise token volume through aggressive pricing.
Rollout begins today across ChatGPT, Codex, and OpenAI's API, with broader access expected over the coming days.
