OpenAI launches GPT-5.6 across ChatGPT, Codex and the API

OpenAI has moved GPT-5.6 from preview into general availability, making the new model family available across ChatGPT, Codex and the OpenAI API. The release is more than a routine model refresh: OpenAI is positioning GPT-5.6 as its new flagship line for agentic coding, professional knowledge work, cyber and science tasks, while also splitting the family into three tiers for different cost and performance needs.

The company says the launch includes GPT-5.6 Sol, its highest-capability model, plus Terra for everyday work and Luna as the lower-cost option. OpenAI's central claim is efficiency: it says GPT-5.6 is designed to produce more useful output per token, with stronger performance per dollar than its previous generation. That is a notable emphasis as businesses continue to weigh the value of advanced AI models against compute budgets and usage caps.

What changes for users and developers

Availability starts immediately, with OpenAI saying the global rollout should continue over a 24-hour period. In ChatGPT, Plus, Pro, Business and Enterprise users can access GPT-5.6 Sol through medium and higher effort settings, while Pro and Enterprise users get an additional Sol Pro option for more difficult tasks. In ChatGPT Work and Codex, Free and Go users get Terra, while paid users can choose among Sol, Terra and Luna.

For developers, OpenAI says Sol, Terra and Luna are available through the API. The Responses API is also gaining Programmatic Tool Calling, which lets GPT-5.6 write and run in-memory programs to coordinate tools and process intermediate results. OpenAI says this can support Zero Data Retention compatibility. A separate multi-agent beta lets the model run concurrent subagents and combine their work inside a single request.

Pricing is now published for the three model sizes. OpenAI lists Sol at $5 per million input tokens and $30 per million output tokens, Terra at $2.50 input and $15 output, and Luna at $1 input and $6 output. Cache writes for GPT-5.6 and later models are billed at 1.25 times the uncached input rate, while cache reads keep a 90% cached-input discount.

Microsoft 365 Copilot is part of the rollout

In a separate announcement, OpenAI said GPT-5.6 will become the preferred model in Microsoft 365 Copilot for Word, Excel, PowerPoint, Chat and Cowork. Microsoft will use OpenAI models natively and also access them through the OpenAI API, according to the announcement. For workplace users, that means the same model family is being pushed into both standalone AI tools and the productivity suites where documents, spreadsheets and presentations are already created.

Sources

• openai.com
• openai.com

Cover photo by Google DeepMind on Pexels, used under the Pexels License.