ChatGPT splits by effort while Google pushes AI further into everyday choices

August 7, 2026

OpenAI changed ChatGPT’s product tiers again. It improved GPT-5.6 Sol for Plus and Pro users, added controls over reasoning effort, and made GPT-5.6 Luna the default model for free users with unlimited text chats and a new Think button for higher-effort questions. Google, meanwhile, expanded Ask Maps in Google Maps so it can order food, book hotels and show ticketed local events, and added a Personal Intelligence option for US users that uses data from Gmail and Google Calendar to tailor recommendations.

This looks like a further split between AI that is priced by how much thinking time a user can call on and AI that is embedded into software people already open for another reason. OpenAI’s change appears to make reasoning effort itself more legible as the thing being sold, rather than access to one undifferentiated chatbot. Google’s move suggests a different pressure on Microsoft: if AI becomes part of navigation, booking and recommendations inside existing consumer products, the contest shifts away from the chat window and towards control of the surrounding service. Taken together, the day’s more consequential moves suggest that generative AI is being sorted less by model labels and more by where it sits in a user’s routine and how explicitly its extra work is metered.