OpenAI Launches GPT-6 Sol and Luna with 50% Cost Reduction and Fewer Errors

OpenAI's GPT-6 Sol and Luna models are now available, offering a 50% cost reduction compared to the GPT-5.6 series. The GPT-6 Sol model has an error rate that is approximately half that of its predecessor, achieving Astra-level reliability. This release follows the earlier introduction of the GPT-6 Astra model earlier this month. OpenAI claims that the new models handle tasks significantly better than Anthropic's top models, including the Fable and Opus series. The internal factuality evaluation conducted by OpenAI indicates that GPT-6 Sol makes about half as many mistakes as previous versions, based on de-identified real-world conversations flagged by users. This advancement allows developers to build more reliable applications at a lower cost, enhancing the overall accessibility of AI capabilities.