OpenAI has unveiled GPT-5.6, its latest family of artificial intelligence models, introducing three new variants called Sol, Terra and Luna. The company says the models deliver improvements in coding, reasoning, agentic workflows and safety, while its flagship GPT-5.6 Sol also outperforms Anthropic’s Claude Fable 5 in front end design.
Alongside the launch, OpenAI said GPT-5.6 Sol achieved an Elo rating of 1353 on Design Arena’s independent front end design leaderboard, ahead of GLM 5.2 at 1351 and Claude Fable 5 at 1345. OpenAI President Greg Brockman described the result as “a big milestone”, pointing to the benchmark as evidence of the company’s progress in software development and web design tasks.
The GPT-5.6 family is designed to serve different use cases. OpenAI describes Sol as its most capable model, built for advanced reasoning, software development and agentic workflows. It also introduces Maximum Reasoning mode for complex problem solving and an Ultra mode that uses sub agents to complete multi step tasks more efficiently.
The company also launched Terra, a balanced model aimed at everyday workloads, and Luna, its lowest cost offering optimised for speed. According to OpenAI, Terra delivers performance comparable to GPT-5.5 at roughly half the cost, while Luna is intended for applications where affordability and response time are priorities.
Beyond coding, OpenAI says GPT-5.6 delivers stronger performance on scientific research benchmarks, including biology, while remaining competitive in cybersecurity evaluations without exceeding its internal safety thresholds. The company also claims the models include stronger safeguards against cyber abuse, sensitive biological misuse and repeated attempts to bypass safety systems.
According to OpenAI, more than 700,000 A100 equivalent GPU hours were used for automated red teaming, alongside extensive testing by human experts before the models were released.
GPT-5.6 is currently available through the OpenAI API and Codex for a limited group of trusted partners, with a broader rollout across ChatGPT planned in the coming weeks. OpenAI has priced Sol at $5 per million input tokens and $30 per million output tokens, while Terra and Luna are positioned as lower cost alternatives.
The launch comes as competition among leading AI companies increasingly shifts beyond chatbot performance towards specialised capabilities such as software development, reasoning and autonomous task execution, where independent benchmarks are becoming an important measure of technical progress.
