Did you open ChatGPT this week and spot model names like Sol, Terra, or Luna? You are not imagining things. On July 9, 2026, OpenAI released GPT-5.6, a new family of three models, and the naming system changed along with it. If the announcement felt like it was written for developers and investors, this post is for you.

Here is a simple explanation of what GPT-5.6 actually is, what the three names mean, which model you get on your plan, and what any of this means for your daily work and study.

What is GPT-5.6?

GPT-5.6 is the newest generation of the AI models that power ChatGPT. According to OpenAI’s official announcement, it launched on July 9, 2026 across ChatGPT, Codex, and the OpenAI API, with a global rollout in the following days.

The headline is not one giant model. It is a family of three: Sol, the flagship; Terra, a balanced model for everyday work; and Luna, the fastest and most affordable one. OpenAI says the whole family was trained to get more useful work out of every token, which in plain language means better answers with less computing, and therefore lower cost. If the word token is new to you, our short guide on what a token in AI means explains it in two minutes.

Sol, Terra, and Luna: what the names mean

The names are Latin for sun, earth, and moon, and they work like sizes rather than versions. The number tells you the generation, and the name tells you the capability tier:

• Sol is the most capable model, built for the hardest tasks.
• Terra is the middle option, tuned for everyday work at a lower cost.
• Luna is the smallest and cheapest, made for speed and volume.

OpenAI says these tiers are here to stay, so future generations should keep the same three names. That is genuinely helpful. Anyone who has tried to explain the difference between older model names to a colleague knows how confusing AI naming has been. If you want a refresher on how these systems work under the hood, see our beginner guide to large language models.

Which model do you get on your plan?

This is the part most readers care about. Based on the official availability notes:

• In ChatGPT, paid users on Plus, Pro, Business, and Enterprise plans get GPT-5.6 Sol. Pro and Enterprise users can also pick Sol Pro for the highest quality on complex tasks.
• In ChatGPT Work and Codex, free and Go users get GPT-5.6 Terra, while paid plans can choose between Sol, Terra, and Luna and set how much effort the model puts in.
• Developers can use all three models through the API.

The interesting detail is that Terra, the model free users get in ChatGPT Work, performs at a level competitive with GPT-5.5, which was the flagship generation before this one. Free access to near flagship quality is a quiet but real win for students and anyone learning AI on a budget.

What is actually new for everyday users?

Beyond benchmark numbers, a few changes matter in practice. OpenAI says GPT-5.6 is noticeably better at producing finished, usable work: editable presentations, well formatted documents, and spreadsheets that follow a template you give it. It is also better at design, so interfaces and visual material it creates need less cleanup. Microsoft clearly agrees, since GPT-5.6 is now the preferred model in Microsoft 365 Copilot.

There is also a new high effort setting called ultra, which runs four AI agents in parallel on one task. It is aimed at heavy professional work and sits behind the higher paid plans, so most casual users will not see it. The takeaway is the direction: AI tools are moving from answering questions toward completing whole pieces of work, something OpenAI leans into with its ChatGPT Work pitch.

From my own experience running websites and testing online tools, launch week claims always deserve a calm head. The demos are impressive, but the honest test is your own work. Give the new model one real task you do every week and compare the result with what you got before.

What about safety?

GPT-5.6 is stronger at cybersecurity tasks than any previous OpenAI model, and that cuts both ways. OpenAI says it shipped its most robust safeguards so far, with layered checks and a system that blocks roughly ten times more potentially harmful activity than before. Its most sensitive security capabilities are reserved for verified professionals through a trusted access program.

For normal users, the practical effect is small. You might occasionally see a refusal on a harmless technical question, and OpenAI offers a retry on a lower capability model when that happens. Working around cybersecurity myself, I would rather see a cautious default than an open door.

Tip: do not pay for the biggest model by default. Pick one weekly task, run it on the model your plan already includes, and only upgrade if the result genuinely falls short.

What this means for you

For most people, GPT-5.6 means the free and cheap tiers of AI just got better, the names finally make sense, and the gap between a chat answer and a finished document keeps shrinking. Prices for builders dropped too, with the smallest model costing one dollar per million input tokens, which keeps pushing capable AI into more of the apps you already use.

It also sharpens the competition. If you are deciding which assistant fits your work, our comparison of ChatGPT, Gemini, and Claude walks through the differences, and our guide to how AI memory works explains what these tools remember about you across chats.

Common questions

Is GPT-5.6 free to use?

Partly. Free and Go users get GPT-5.6 Terra in ChatGPT Work and Codex, which OpenAI says performs at a level competitive with the previous flagship generation. The top model, Sol, needs a paid plan.

What do Sol, Terra, and Luna mean?

They are capability tiers named after the sun, earth, and moon. Sol is the most powerful, Terra is the balanced middle option, and Luna is the fastest and cheapest. The number, 5.6, is the generation.

Is GPT-5.6 better than GPT-5.5?

On OpenAI’s published evaluations, yes, across coding, knowledge work, and science, and it usually gets there faster and with fewer tokens. As always, benchmark wins do not remove the need to check important facts in anything an AI writes for you.

Final takeaway

GPT-5.6 is less about one dramatic breakthrough and more about better results for less money, with clearer names on the box. Try the model your plan already gives you on a real task this week, keep your usual habit of checking its output, and you will get most of the benefit without spending anything extra.