OpenAI shipped the first major update to GPT-5.6 since the model family launched. The Sol tier — available to ChatGPT Plus and Pro subscribers — gets a real accuracy and focus improvement, and the free tier becomes materially more useful: free users are now defaulted to GPT-5.6 Luna with unlimited text chat.
Two changes stand out. First, OpenAI added a slider that lets you control how much effort the model puts into thinking on each response. Second, free users get a Think button that routes harder questions to deeper reasoning. That combination moves the free tier from “useful for casual chat” toward “useful for actual work,” which is a meaningful shift for anyone who has been living on the free plan.
What Changed for Plus and Pro Users
The Sol update targets two things users complained about most: factual accuracy and answer focus. OpenAI says the improved Sol is more accurate on factual questions and stays tighter on what you actually asked, less rambling, fewer irrelevant tangents. In practice, that is the difference between a model that answers and a model that writes around the answer.
The new thinking-effort slider is the more interesting change. Instead of a fixed level of reasoning, you dial how much the model should think before responding. Low effort for quick factual lookups and simple rewrites; high effort for math, code, and multi-step problems. If you have been paying for Sol but feeling like you were getting the same output regardless of prompt complexity, this slider is the missing control.
What Changed for Free Users
The free plan now defaults to GPT-5.6 Luna, and OpenAI lifted the previous limits: free users get unlimited text chat instead of a capped number of messages. The Think button is the deliberate bridge to the paid tier, when a question needs deeper reasoning, one tap routes it to the harder model.
This matters for two groups. Students and light users get a genuinely more capable free tier for everyday work. And anyone evaluating whether to subscribe now has a cleaner test: use Luna free for as long as it covers your needs, and upgrade only when you regularly hit the wall where thinking depth matters.
The Strategic Read
OpenAI is clearly trying to widen the funnel. Unlimited chat on the free tier is a competitive answer to the equally generous free tiers from Google and Anthropic, and the Think button is a conversion tool disguised as a feature. The message to users is honest, though: the gap between free and paid is no longer about volume of chat. It is about depth of reasoning, which is where the money is.
The Caveats
Two things to keep in mind. First, “unlimited” applies to standard text chat, heavy features and high-intensity usage can still be throttled, and OpenAI can revise terms quickly. Second, the Sol improvements are incremental, not revolutionary; if you already find Sol’s accuracy acceptable, the update is a polish pass, not a reason to re-evaluate the product. Check the official announcement for the current terms before you change anything about how you use ChatGPT.
Related: see our ChatGPT vs Claude comparison