OpenAI has outlined details of its upcoming GPT-5.6 artificial intelligence models, introducing a three-tier lineup designed to balance performance, speed and cost efficiency as competition intensifies in the AI sector.
The company said the GPT-5.6 family will consist of three models — Sol, Terra and Luna — which are scheduled to roll out over the coming weeks.
GPT-5.6 Sol will serve as OpenAI’s flagship model, targeting advanced applications in cybersecurity, biological sciences and software development. According to the company, the model delivers improved performance compared with GPT-5.5 while reducing token consumption, helping users achieve greater efficiency.
OpenAI said Sol will also feature new “Max” and “Ultra” modes that allow deeper reasoning capabilities and the use of multiple AI agents for more complex tasks.
The mid-tier GPT-5.6 Terra model is designed to strike a balance between performance and speed. OpenAI described Terra as offering capabilities similar to GPT-5.5 while operating at less than half the cost, making it suitable for a broader range of commercial and enterprise applications.
Meanwhile, GPT-5.6 Luna has been developed with efficiency as its primary focus. OpenAI said Luna maintains strong capabilities while costing more than 50% less than Terra, positioning it as the most affordable option in the new lineup.
The announcement comes as scrutiny over advanced AI systems continues to grow, with governments and regulators increasingly focused on safety and security concerns surrounding powerful generative AI models.
OpenAI said all three GPT-5.6 models include multiple layers of safeguards designed to prevent misuse. The company highlighted Sol’s ability to identify software vulnerabilities and assist in developing security fixes while resisting attempts to generate complete exploit chains that could be used by malicious actors.
However, OpenAI acknowledged that the safety measures in the initial GPT-5.6 preview may be overly restrictive in some cases, potentially limiting legitimate use scenarios. The company said it plans to refine those protections over time based on user feedback.
The GPT-5.6 preview will initially be made available to selected trusted partners and organisations. Broader access through ChatGPT and Codex is expected at a later stage.
The launch marks OpenAI’s latest effort to expand its AI portfolio as demand for advanced language models continues to accelerate across industries.
