OpenAI launches GPT-6 Sol and Luna with better performance and 50% lower API prices

OpenAI has released GPT-6 Sol and GPT-6 Luna, new members of the GPT-6 family. They use a training approach similar to GPT-6 Astra, bringing Astra's advanced performance in professional work, factual accuracy, coding, and computer use to faster and cheaper models. The API price for both models is 50% lower than the GPT-5.6 promotional price.

In tests, GPT-6 Sol outperformed Claude Opus 5 on the AutomationBench cross-app workflow test, at only 9% of Opus 5's cost per task. On the Last Exam agent evaluation, Sol scored 56.4%, higher than Opus 5's best score, with 60% lower cost per task. For coding, Sol scored 68.8% on DeepSWE v1.1, close to Claude Fable 5's 69.9%, while costing about 80% less per task.

Both models are now available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users. Free and Go users can use GPT-6 Luna in the desktop app. They are available in the API as gpt-6-sol and gpt-6-luna.