Which ChatGPT Model Should You Use in 2026? (GPT-6 Instant, Medium, High and Pro Explained)

Short answer (updated November 2026): On a paid plan, pick GPT-6 in the model menu and leave it on Instant for everyday work. Move the Thinking slider to Medium or High for hard problems like math, coding and planning. Use Pro only for the hardest tasks, if your plan has it. On the Free plan you don't pick a model. You get OpenAI's fast Luna model, with a Think button for harder questions.

The rest of this guide shows which option to use for which task, what each plan includes, and what the new names mean. Every fact below was checked against OpenAI's help center and the ChatGPT pricing page in November 2026. Model names change often, so check the model menu in your own account too.

Which ChatGPT Model Should I Use? Quick Table

Your task	Pick this	Plan you need
Everyday questions, emails, summaries, rewriting	GPT-6 Instant	Plus or higher (Free and Go get Luna)
Writing a blog post, essay draft or social post	GPT-6 Instant	Plus or higher
Homework help with several steps, study plans	GPT-6 Medium	Plus or higher (Free and Go: Think)
Hard coding bugs, math, data analysis	GPT-6 High	Plus or higher
Long, careful reasoning on a hard problem	GPT-6 Extra High	Pro, Business or Enterprise
Research-level problems, long-running work	Pro (GPT-6 Pro, powered by GPT-6 Astra)	Pro, Business or Enterprise
You want to stay on the older model	GPT-5.6 Sol	Plus or higher

A simple rule: start on Instant. If the answer is wrong or shallow, ask again one level up.

How the ChatGPT Model Menu Works Now

ChatGPT no longer asks you to choose between many model names in a chat. On paid plans you choose GPT-6, then set how much it thinks with the Thinking slider. GPT-6 replaced the old "Latest" option in the menu.

Here is what each level does, in OpenAI's own words, summarized:

• Instant: fast answers for everyday questions.
• Medium: standard thinking.
• High: extended thinking.
• Extra High: the highest thinking level available with GPT-6 Sol.
• Pro: Pro models for difficult tasks and longer-running work.

Behind the scenes, GPT-6 Sol runs Instant, Medium, High and Extra High. The Pro level runs GPT-6 Pro, which is powered by GPT-6 Astra. You won't see "Sol" in the menu. It is labeled GPT-6.

More thinking gives better answers on hard problems, but it is slower and uses up your Thinking limit faster. For easy tasks it adds wait time and nothing else.

Instant vs Medium vs High: which one?

• Instant for anything you would ask a smart friend: explain this, rewrite this, summarize this, give me ideas.
• Medium when the task has steps that depend on each other: a study plan, a budget, a short program, a tricky word problem.
• High when Medium gets it wrong, or when a mistake would cost you time: debugging, proofs, analysis you will hand in or rely on.

If you are not sure, try Instant first. It costs you a few seconds to find out.

What the Free Plan Gets

Free users don't have the GPT-6 menu. Here is what you get:

• Unlimited everyday text chats, within OpenAI's abuse safeguards. Uploads, image creation, voice and other tools still have limits.
• Luna as your model. Luna is the fastest, lowest-cost model in OpenAI's lineup. The help center names GPT-5.6 Luna as the default for Free and Go. The pricing page already lists GPT-6 Luna for every plan, so you may see the newer version as it rolls out.
• Think for harder questions. Think also runs on Luna, not on the bigger Sol model.
• GPT-5 Thinking Mini, a small reasoning model, is listed for Free on the pricing page.
• No GPT-6 Sol, GPT-5.6 Sol or Pro models.

The Go plan gets the same models as Free, with higher limits on tools, uploads and memory.

For most students, Free is enough for daily study. Upgrade when you keep needing deeper reasoning than Think gives you. Our ChatGPT Free vs Plus vs Pro guide compares the plans in full.

Which Plan Gets Which Model

Model	Free	Go	Plus	Pro
Luna (everyday chat and Think)	Yes	Yes	Yes	Yes
GPT-6 Instant, Medium, High	No	No	Yes	Yes
GPT-6 Extra High	No	No	No	Yes
GPT-6 Pro (Astra) in Chat	No	No	No	Yes
GPT-5.6 Sol	No	No	Yes	Yes
GPT-5.6 Sol Pro	No	No	No	Yes
Older (legacy) models	No	No	Yes	Yes

Business and Enterprise workspaces get Medium, High, Extra High and Pro, but an admin can limit which models members see. Plus users can use GPT-6 Astra in ChatGPT Work and Codex, just not in normal chats.

Context window sizes (the total amount ChatGPT can keep in view) also differ by plan. For instant answers it is about 27K tokens on Free, 54K on Go and Plus, and 128K on Pro. For reasoning it is 256K on Go and Plus and 400K on Pro.

The Other Names You May See

OpenAI uses a lot of model names. These are the ones that cause the most confusion:

• GPT-5.6 Sol is the model before GPT-6. It is still in the menu, and if you picked it on purpose, your chats stay on it. GPT-6 did not retire it.
• GPT-5.6 Sol Pro is the Pro option of the GPT-5.6 family, for Pro plans.
• GPT-5.6 Terra is a balance of skill, speed and cost. You can't pick it in a normal chat. It is used in ChatGPT Work, Codex and the API.
• GPT-6.1 Sol and GPT-6.1 Terra are model options in ChatGPT Work and Codex only. They are not in the normal chat menu.
• GPT-6 Astra is the model behind GPT-6 Pro. You meet it as the "Pro" level, not as its own menu item.

So for normal chatting, the choice is short: GPT-6 and a thinking level, or GPT-5.6 Sol if you want the older model.

Usage Limits to Know

• Everyday text chat is unlimited on every plan, Free included.
• Medium, High and Extra High count against a Thinking limit. If you hit it, ChatGPT may keep going with another available model. ChatGPT shows when your limit resets when it can.
• GPT-6 Pro is not unlimited, even on Pro plans. On Pro $200, when the weekly GPT-6 Pro limit runs out, ChatGPT switches to GPT-5.6 Thinking at Medium.
• OpenAI Support can't reset your limits. You wait for the reset or use another model.

OpenAI does not publish a single fixed message number for every plan, and the numbers change. Check the reset time shown in your account.

Practical Examples

• Drafting a blog post or cover letter: Instant. More thinking won't make the writing better, only slower.
• Summarizing a 20-page PDF: Instant. On Free, very long files may not fit in the smaller context window.
• Making a two-week exam study plan: Medium.
• Fixing a bug you have stared at for an hour: High.
• Solving a hard math problem step by step: High, then Extra High or Pro if your plan has them and High gets it wrong.
• Writing 10 headline ideas: Instant.
• Free plan, tricky homework question: turn on Think.

Use the Right Model, Not the Biggest One

Most of what people do in ChatGPT is a language task: writing, explaining, summarizing. Instant handles those well and fast. Thinking levels are for problems with many steps, where checking the work matters.

Ask yourself one question before you pick: do I need words, or do I need a worked-out answer? Words, use Instant. A worked-out answer, move the slider up. To learn why thinking models behave differently, read reasoning models vs regular models.

Once you have the model right, the next step is better prompts. Our free ChatGPT for Complete Beginners course gets you started, and the free ChatGPT Power User course covers model choice, prompting and workflows. For hands-on prompt practice, try our free Prompt Engineering course. Each one comes with a free certificate.

If you are picking a model to power business automation instead of personal use, Cynked's 7 AI agents every e-commerce business should deploy in 2026 is a useful frame for matching the model to the job.

Frequently Asked Questions

What is the latest ChatGPT model in 2026?

GPT-6. It is rolling out as the default in chats for Plus, Pro, Business and Enterprise users. GPT-6 Sol runs the Instant to Extra High levels, and GPT-6 Pro, powered by GPT-6 Astra, runs the Pro level. Free and Go users get Luna instead.

Which ChatGPT model is best?

For most tasks, GPT-6 at Instant. For hard problems, move up to Medium or High. Pro (GPT-6 Astra) is the strongest option, but it is slower, has its own usage limit, and is only on Pro, Business and Enterprise plans.

What is the difference between Instant, Medium and High in ChatGPT?

They are thinking levels for the same model. Instant answers fast. Medium thinks before answering. High thinks for longer. More thinking helps with math, code and multi-step problems, but it is slower and uses your Thinking limit.

Which ChatGPT model do free users get?

Free users get Luna, OpenAI's fastest and lowest-cost model, with unlimited everyday text chats. The help center names GPT-5.6 Luna as the default, and the pricing page lists GPT-6 Luna. Free users can press Think for harder questions, which also runs on Luna.

Does ChatGPT Plus include GPT-6 Pro?

Not in normal chats. Plus gets GPT-6 at Instant, Medium and High, plus GPT-5.6 Sol. GPT-6 Pro in chats is for Pro, Business and Enterprise plans. Plus users can use GPT-6 Astra in ChatGPT Work and Codex.

Can I still use GPT-5.6 in ChatGPT?

Yes, on paid plans. GPT-5.6 Sol is still in the model menu, and chats where you picked it on purpose stay on it. GPT-5.6 Terra can't be picked in a normal chat. It is used in ChatGPT Work, Codex and the API. GPT-5.6 Luna is the automatic model on Free and Go, not something you pick from the menu.

What are GPT-6.1 Sol and GPT-6.1 Terra?

They are model options in ChatGPT Work and Codex. They are not in the normal chat menu, so most people choosing a model for everyday chats can ignore them.

---

Published: April 1, 2026. Last updated: November 2, 2026 for GPT-6 in ChatGPT, the new Thinking levels (Instant, Medium, High, Extra High, Pro) and the models each plan gets. Sources: OpenAI's help center article "GPT-6 and other models in ChatGPT" and the ChatGPT pricing page.