Back to the stories

OpenAI rolls out GPT-6 Sol and Luna across ChatGPT plans

Score 10.0,

Frontier model

The model powering ChatGPT conversations just moved to a new generation, and that change now shows up for most users by default.

OpenAI is rolling GPT-6 out across its ChatGPT plans. Paid tiers (Plus, Pro, Business, Enterprise) will use GPT-6 Sol as the default, and Free and Go users will default to GPT-6 Luna shortly after.

Why this matters: this is not a niche upgrade behind a paywall. OpenAI says GPT-6 is built to serve the platform’s user base at scale, and the company claims it can start giving a useful answer before it finishes coming up with a fuller, better one. In plain terms, conversations get a quick, helpful reply almost immediately and then gain more depth as the model keeps reasoning in the background.

How it works, simply put: imagine a chef who brings a plated appetizer to the table while still finishing the main course. The model delivers a first pass fast, then continues refining and expanding that answer. OpenAI also says its top “Extra High” setting now reaches a higher quality than the previous generation while responding about as quickly as mid-tier responses used to.

What changes now: products and teams that embed ChatGPT get that quality and speed improvement without asking for a new model key or separate access. The practical limit is unchanged, running this kind of frontier model yourself still requires massive compute, so for now the upgrade matters most through OpenAI’s hosted service.

The open question is straightforward: will independent benchmarks and real-world use confirm both the faster responses and the deeper follow-up reasoning? That will determine whether this is an evolutionary polish or a genuine step forward for everyday AI.