OpenAI Launches GPT-6 Sol and Luna with Improved Efficiency and Lower Costs
OpenAI has launched updated versions of its smaller AI models, GPT-6 Sol and Luna. These new iterations aim to make the intelligence introduced by the more powerful GPT-6 Astra model more efficient and accessible for a broader range of tasks. Sol is tailored for complex work such as coding, while Luna targets high-volume clerical tasks like document summarization or information extraction. OpenAI highlights significant improvements in accuracy and efficiency, with Sol making about half as many mistakes compared to its predecessor on internal evaluations. The models are also available at a 50% lower cost than the previous versions due to advancements in caching and inference. The release comes amid fierce competition between AI leaders like Anthropic, who recently updated their Opus model just hours before OpenAI’s announcement. GPT-6 Sol and Luna will be rolled out across various platforms including ChatGPT Work, Codex, and the API.
Topics: artificial intelligence GPT-6 Sol and Luna model launch OpenAI