Originally announced on June 26 and with a release teased on Wednesday, the latest and greatest ChatGPT model family has arrived. 

ChatGPT-5.6 Sol, Terra and Luna are available starting today in ChatGPT, Codex and the OpenAI API, OpenAI said in a blog post. The model family is currently rolling out globally and should be available to all users “within the next 24 hours.” 

OpenAI has been at the forefront of the artificial intelligence movement since releasing ChatGPT nearly four years ago, and it’s kept its lead in terms of overall users, despite increasing competition from rivals like Google and Anthropic. 

(Disclosure: Ziff Davis, CNET’s parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

Sol is the new flagship model and is built to handle complex reasoning, end-to-end technical execution and design. OpenAI calls it the company’s best coding model yet, scoring better than Anthropic’s Fable 5 model in the Artificial Analysis Coding Agent Index. Benchmarks show it’s also more efficient across token output, time and cost. 

Terra, the middle-of-the-road model, is intended for everyday workflows and may be the go-to option for most people. Luna prioritizes cost efficiency and speed. 

OpenAI said ChatGPT-5.6 is even better at design now and can inspect the rendered result to catch visual issues. It also touted 5.6 as its strongest cybersecurity model yet, achieving high performance with fewer tokens. 

The cybersecurity capabilities of the latest frontier models — GPT-5.6 and Anthropic’s Mythos 5 and Fable 5 — are why the companies have been slower to make them available to the general public. In Anthropic’s case, the company first made a version available only to select users for cybersecurity purposes. When it brought a version to the public, the Trump administration ordered the company to prohibit any foreign national from using it, effectively banning it while the government reviewed its capabilities and took safety precautions. GPT-5.6 was similarly first released to cybersecurity and technology professionals, with a staggered release requested by government officials.

ChatGPT Work

The ChatGPT-5.6 models aren’t the only announcement OpenAI had up its sleeves on Thursday, as it also launched ChatGPT Work. It’s described as an agent that lives within ChatGPT that can take action within your apps and files. 

ChatGPT Work is powered by the new 5.6 models, giving it strong reasoning skills and allowing it to gather information across your apps, create a plan and execute it with little to no interaction on your part. You can also keep the agent at work with Scheduled Tasks while away from your computer.