
OpenAI is removing rate limits on text conversations for free-tier users this week while transitioning them to the new GPT-5.6 Luna model.
Internal testing showed the new Luna and Sol models reduced factual errors by 62 percent and 68 percent respectively compared to the previous GPT-5.5 Instant model.
Plus and Pro subscribers now access the updated Sol model, which features a new slider for users to adjust the reasoning effort applied to responses.