Choosing Models and Reasoning Effort in ChatGPT: GPT-6 Sol, Astra, Luna and Terra
The rule for choosing a model in ChatGPT is to start with the default effort and raise it only for multi-step work with trade-offs; the Thinking setting in Chat and the Power presets in Work and Codex are separate controls. The GPT-6 menu as of 15 November 2026, which jobs Astra, GPT-6.1 Sol, Terra and Luna suit, Max and Ultra, the GPT-5.5 retirement, and a method for comparing a Bookstore campaign analysis at two settings.
15 Nov 2026
·
11 min
·
6
