Tech Desk: OpenAI has just rolled out GPT-6 Sol and Luna, a significant expansion of its GPT-6 generation that promises to make advanced AI more accessible and reliable. Following the powerful GPT-6 Astra, these updated models are set to revolutionize how businesses and individuals leverage artificial intelligence, boasting a remarkable 50% price reduction for API access compared to their predecessors. This cost efficiency, stemming from advancements in caching and inference, makes cutting-edge AI capabilities more affordable than ever.

Beyond affordability, accuracy is a cornerstone of this release. OpenAI claims GPT-6 Sol now makes about half as many mistakes as its previous version, achieving Astra-level reliability, especially in complex tasks like coding. Luna, designed for high-volume clerical work such as summarizing documents or extracting information, also exhibits enhanced precision, minimizing errors across a range of applications.

The launch intensifies the fierce competition in the AI landscape, arriving shortly after Anthropic’s Opus 5.5 release. OpenAI explicitly positions Sol and Luna as superior to Anthropic’s top models, Fable and Opus, in handling various tasks. This direct comparison underscores the rapid innovation and competitive drive within the industry.

Currently, the new versions of Sol and Luna are rolling out across ChatGPT Work, Codex, and the ChatGPT API for paid accounts, with Luna also becoming available for Free and Go users via the desktop app. This phased deployment ensures that a broader audience can soon experience these more efficient and accurate AI models.