OpenAI begins GPT-5.6 rollout with Sol, Terra and Luna models
OpenAI has introduced the GPT-5.6 family, featuring Sol, Terra and Luna models designed for advanced reasoning, coding and everyday AI tasks.
OpenAI is taking a different approach to its next generation of AI. Instead of launching a single model for every user, the company has introduced the GPT-5.6 family, a tiered lineup consisting of Sol, Terra and Luna.
Each model has been designed for different workloads, allowing developers and organisations to choose the right balance of performance, speed and cost. The rollout began as a limited preview on 9 July 2026, with access currently restricted to selected partners through the OpenAI API and Codex.
Meet Sol, Terra and Luna
At the top of the lineup is GPT-5.6 Sol, OpenAI's most powerful model in the preview. Sol is built for demanding tasks such as software engineering, scientific research, advanced reasoning and complex problem-solving.
According to OpenAI, it delivers improvements across professional knowledge work, cybersecurity, computer use and technical analysis. GPT-5.6 Terra serves as the middle-tier option. It is designed for everyday productivity, including writing, document analysis, coding assistance and brainstorming.
The AI firm says Terra offers performance comparable to GPT-5.5 while costing around half as much, making it suitable for organisations that want strong AI capabilities at a lower price.
GPT-5.6 Luna is the fastest and most affordable model in the family. It focuses on lightweight tasks such as conversations, summaries and routine assistance where speed and efficiency are priorities.
Stronger reasoning and coding capabilities
One of the biggest improvements in GPT-5.6 is enhanced reasoning. Reasoning allows an AI model to spend more time analysing complex questions before generating an answer. This leads to better performance on tasks involving multiple steps, detailed instructions and logical problem-solving.
For developers, GPT-5.6 can assist with reviewing code, identifying bugs and suggesting improvements. Business users may also benefit from more reliable document drafting, spreadsheet analysis, project planning and research support.
Advances in safety measures
Alongside its performance improvements, OpenAI says GPT-5.6 Sol introduces its most advanced safety protections to date. According to the company, the models include stronger defences against high-risk requests, cybersecurity misuse and repeated attempts to exploit the system.
OpenAI also says GPT-5.6 Sol underwent several weeks of testing before entering preview to improve its resilience against real-world threats.
When will GPT-5.6 be available?
For now, GPT-5.6 remains in a controlled preview phase. Access is limited to selected organisations using the OpenAI API and Codex, and the new models are not yet available in ChatGPT. OpenAI has not announced when a wider public release will take place.
The GPT-5.6 family also signals a broader shift in AI development. Rather than relying on one model for every scenario, OpenAI is creating specialised options that allow users to match capability, speed and cost to specific tasks. If the preview proves successful, this tiered approach could shape how future AI models are delivered to businesses and developers.