The models behind ChatGPT right now (GPT-6 Luna, Sol and Astra), which plan gets which, and when to use each thinking level. Updated monthly.
What it is
ChatGPT is powered by AI models. A model is the "brain" that writes the answer. OpenAI releases new models every few months, and the model you get depends on your plan and the choice you make in the model menu.
This is the one page we update every month. Other lessons on this site talk about features, not model numbers, so they stay correct when models change.
Checked: October 29, 2026.
The current models: the GPT-6 family
On September 22, 2026, OpenAI launched the GPT-6 family. In October it removed GPT-5.5 from ChatGPT, so GPT-6 models now power most chats.
| Model | What it is good at | Who gets it in ChatGPT |
|---|---|---|
| GPT-6 Luna | Fast answers for everyday jobs: quick questions, summaries, pulling out information | Free and Go users |
| GPT-6 Sol | OpenAI's main model for harder work, such as coding, analysis and long writing | Plus, Pro and Business |
| GPT-6 Astra | The largest and most capable model, used for the Pro setting | Pro and Business (when the admin allows it) |
Thinking levels
On paid plans you can choose how long the model "thinks" before it answers:
- Instant: answers right away. Best for most everyday questions.
- Medium / High / Extra High: the model reasons step by step before answering. Slower, but better for maths, code, planning and tricky questions.
- Pro: the deepest setting, for the hardest problems. Available on Pro and Business.
Free and Go users can still ask ChatGPT to think harder on some messages, with tighter limits.
How to choose a model
- Click the model menu at the top left of the chat.
- Pick a model or thinking level.
- Send your message. You can change the level for each new message.
Example
Sana asks "What's a good name for a bakery?" with Instant and gets ten ideas in two seconds. Then she asks for a 12-month cash-flow plan for the bakery. For that she switches to High thinking. It takes longer, but the numbers add up and each assumption is explained.
Try it yourself
Send the same prompt twice, once with Instant and once with a higher thinking level (if your plan has it), and compare:
A shop sells pens at 3 for ₹50. How much do 14 pens cost if any extra pens are sold at the single-pen price? Show your working.
What happens when you hit a limit
Every plan has usage limits that reset after a few hours. When you reach yours, ChatGPT tells you and switches to a smaller model until the limit resets. Your chat keeps working, but answers to hard questions may be less detailed.
Tips and common mistakes
- Don't use the strongest setting for everything. It is slower and uses up your limit faster. Use Instant for simple jobs.
- Use thinking for maths, code and multi-step plans. These are where reasoning makes the biggest difference.
- Names change fast. If your menu shows different names from this page, trust your screen and OpenAI's help pages. We update this page monthly.
Free vs paid
Free and Go use the fast everyday model with limits. Plus unlocks the main model and thinking levels. Pro adds the Pro setting and the highest limits. See ChatGPT Plans Explained.
Next: write better requests in Your First ChatGPT Prompt.
FAQ
What model does ChatGPT use now?
As of late October 2026, ChatGPT uses OpenAI's GPT-6 family. Free and Go users get GPT-6 Luna, paid plans use GPT-6 Sol, and the Pro setting is powered by GPT-6 Astra.
What is the difference between Instant and Thinking in ChatGPT?
Instant answers right away and suits everyday questions. Thinking levels make the model reason step by step first, which is slower but more accurate for maths, code and planning.
Is GPT-5.5 still available in ChatGPT?
No. OpenAI removed GPT-5.5 from ChatGPT in October 2026. It remains available to developers through the API.
What happens when I reach my ChatGPT limit?
ChatGPT switches you to a smaller model until your limit resets after a few hours. You can keep chatting in the meantime.