That means users on the Free plan get updated to GPT-5.6 Luna as the standard, with unlimited text chats. They can also access a better reasoning mode with a new «Think» button that can be used for tougher queries.
For Plus and Pro users, Sol is getting promoted to the default, even in the standard chat mode, Instant, that used to be handled by GPT-5.5.
At the same time, OpenAI says they have updated the Sol model itself. It should now deliver «more focused answers,» yet no news on the hedges, caveats and nitpicking that some users find annoying. In fact, OpenAI says it now «offers a helpful correction when simply agreeing wouldn’t be useful.» GPT Sol should also be 68% less error prone than GPT-5.5 Instant.
In addition, Plus and Pro users get an updated slider that pops up in the chat bar to immediately adjust the reasoning level, from Instant to High, Extra High and Pro, depending on the subscription.
The updates should be available «now,» but might take some time to propagate over the internet. The Sol version in use for Work and Codex won’t be changing.
Read more: OpenAI’s presentation, X post. Writeups on TechCrunch, Axios, and The Verge. Discussions on Hacker News and r/Singularity.