GPT-5.6 Sol unifies paid ChatGPT modes as Luna opens free text

ChatGPT’s paid model experience is being consolidated around GPT-5.6 Sol. In an August 6 X post, OpenAI said Sol now powers both Instant and deep reasoning for Plus and Pro users, while Free and Go users will get unlimited GPT-5.6 Luna text chats starting the next day.

OpenAI wrote: “GPT-5.6 Sol now powers both Instant and deep reasoning.”

The concrete number is 68%. In a follow-up post in the same product thread, OpenAI said GPT-5.6 Sol produced 68% fewer responses with factual errors than GPT-5.5 Instant in a high-stakes factuality evaluation spanning finance, medicine, and law. That frames the rollout less as a naming change and more as an attempt to make the default ChatGPT path more reliable in domains where errors have real cost.

The OpenAI account is the company’s direct channel for ChatGPT model rollouts, access changes, and safety or evaluation notes. This post did not link to a full technical report; it used a short video and compact copy to state the product behavior. That means the public evidence is strong on availability and positioning, but thinner on benchmark methodology, test size, and whether the same 68% gain carries into everyday chat traffic.

For users, the immediate shift is practical: paid tiers no longer have the same split between fast chat and deeper reasoning, and free users get a broader text allowance through Luna. The next things to watch are the model picker inside ChatGPT, whether Sol and Luna appear in API-facing docs, and whether OpenAI publishes the evaluation design behind the 68% factual-error reduction.

Source: OpenAI on X