OpenAI: Sol and Luna Series 6 at Half Price with Expanded Access

OpenAI is rolling out new versions of Sol and Luna in its professional offerings and via the API, with deployment announced starting today. The Series 6 models are arriving at half the price of Series 5.6, while the company claims fewer errors and improved factual accuracy. This launch comes shortly after an update from Anthropic, signaling tight competition.

Availability Expands to Professional Offers and API Starting Today

The new versions of Sol and Luna are available in ChatGPT Work and Codex for most paid accounts, and they are accessible via the ChatGPT API. OpenAI announces a gradual rollout in ChatGPT, on the app and website, throughout the day. Luna is also set to join the desktop app and be offered to Free and Go users. This timeline comes as Anthropic released Opus 5.5 just ninety minutes before OpenAI's launch.

Prices Halved and Claimed Efficiency Gains

OpenAI states that the Series 6 models will be offered at half the price of the Series 5.6 versions of Sol and Luna, a reduction the company attributes to improvements in caching and inference. It also claims advancements in factual accuracy and a reduced error rate in coding. In an internal evaluation based on anonymized real conversations, GPT-6 Sol reportedly made about half as many errors as its predecessor, with reliability presented as equivalent to that of Astra at a lower cost.

Sol for Code, Luna for Administrative Tasks, and the Range Aligns with Astra

OpenAI says it is expanding GPT-6 generation with updated versions of its more compact models, Sol and Luna, introduced earlier this year and positioned at distinct levels within its hierarchy. Sol is intended for complex tasks, including coding, while Luna is presented as suitable for high-volume administrative tasks, such as document synthesis, information extraction, or quick responses. In this context, GPT-6 Astra, launched earlier this month, is described by OpenAI as its most powerful model and the best in the world for various activities. The company further asserts that Sol and Luna outperform Anthropic's Fable and Opus models.