OpenAI launches GPT-6 Sol and Luna with lower costs and fewer mistakes

OpenAI has launched updated versions of its GPT-6 Sol and Luna models, expanding the GPT-6 generation that also includes GPT-6 Astra. The company says the new models extend Astra’s capabilities with greater efficiency and accessibility.

Sol is designed for complex tasks such as coding. OpenAI says Luna is intended for clerical work with clear goals, including summarizing documents, extracting information, and answering quick questions.

OpenAI says the GPT-6 models will cost half as much to access through its API as the 5.6 series of Sol and Luna. The company attributes the reduction to improvements in caching and inference.

The company also claims that GPT-6 Sol makes about half as many mistakes as its predecessor in an internal factuality evaluation based on de-identified real-world conversations in which users reported errors. OpenAI says this gives Sol Astra-level reliability at a lower cost. These are company claims.

OpenAI further claims that GPT-6 Sol and Luna perform substantially better than Anthropic’s leading models, including Fable and Opus. Anthropic released Opus 5.5 90 minutes before OpenAI’s announcement.

The new Sol and Luna versions are available in ChatGPT Work and Codex for most paid accounts, and through the ChatGPT API. Luna is also available in the desktop app and for Free and Go users. OpenAI expects a gradual rollout to ChatGPT throughout the day.