OpenAI releases GPT-6 Sol and Luna with lower prices and fewer errors

In short: OpenAI has launched two new AI models, GPT-6 Sol and GPT-6 Luna, and says they are cheaper to use and make fewer mistakes.

What happened

OpenAI announced updated versions of its smaller GPT-6 models, called Sol and Luna. The company says they are based on the same newer “generation” as GPT-6 Astra, which OpenAI released earlier this month.

OpenAI positions the two models for different kinds of work. It says Sol is meant for more complex tasks like coding. It says Luna is meant for high-volume office tasks with clear goals, like summarizing documents, pulling key details from text, and answering quick questions.

OpenAI also said it cut the price for API access to these models by about half compared with the earlier Sol and Luna 5.6 series. An API is a way for other apps and services to “plug in” to OpenAI’s models (like using a power outlet instead of building your own generator). OpenAI linked the lower price to internal efficiency changes, including caching and inference, which are ways to reuse work and produce answers with less computing effort.

The company also claimed the new models are more accurate on facts and make fewer coding errors. It said an internal test based on real user chats where mistakes were flagged showed GPT-6 Sol makes about half as many factual mistakes as the previous version.

Why it matters

For regular users and businesses, lower prices can mean more AI features in everyday tools, and fewer errors can mean less time spent double-checking results. OpenAI is also rolling these models out across ChatGPT products and developer access, so people may notice changes quickly.

Source: TechCrunch AI