Free can use GPT-6 for everyday text chat. That is not GPT-6 Sol in the Plus model menu, and it is not GPT-6 Astra on Pro. Plus at $20 a month buys Sol at Instant, Medium, and High. Pro, from $100 a month, adds Extra High and the Pro thinking level inside Chat. Intelligent UI runs through Extra High. Move the slider to Pro and the model becomes Astra, which does not render this interface.
Today is 16 October 2026. The rollout dates come from the 7 October product page. Thinking levels, interface limits, and weekly allowances come from the Help Center articles on models and Intelligent UI. Context windows come from the pricing page opened the same day. For how to ask for a chart or a calculator, see the 8 October article. This one only answers whether to pay, and where the interface is absent.
Check the model menu first
Three official pages do not say the same sentence. Before you pay, open the model menu in Chat.
- The 7 October product page. That day GPT-6 and Intelligent UI started rolling out in Chat to Plus, Pro, Business, and Enterprise. Free and Go followed on 8 October. Paid everyday chat uses GPT-6 Sol. Free and Go use GPT-6 Luna. The release changes Chat only. Work and Codex do not switch models because of it. Enterprise also depends on an admin.
- The pricing page. The first Free bullet is unlimited GPT-6 text chats, still subject to abuse guardrails. In the comparison table, GPT-6 Sol and GPT-6 Astra are absent on Free and Go and present on Plus. GPT-6 Luna is present on Go and Plus.
- The Help Center model article. Free and Go do not include Medium, High, Extra High, or Pro. The same page says Instant and Think on those two plans use GPT-5.6 Luna, not GPT-5.6 Sol. If GPT-6 is missing from a paid account’s menu, that account has not received it yet.
If the menu already says GPT-6, Free and Go map to Luna and Plus and above map to Sol. If the menu still says GPT-5.6 Luna, you are on the tier the Help Center describes. Neither case includes Plus-level Medium and High. A different prompt will not turn Luna into Sol.
What Free, Go, Plus, and Pro actually get
Which model runs everyday chat
| Plan | Monthly price (USD) | Everyday model in Chat | Pro thinking in Chat |
|---|---|---|---|
| Free | 0 | Product page: GPT-6 Luna. The Help Center also says Instant may still be GPT-5.6 Luna | No |
| Go | 8 (US) | Same Luna tier. Some countries use a local price | No |
| Plus | 20 | GPT-6 Sol | No. Astra is in Work and Codex, not on this slider |
| Pro 100 / 200 / 500 | 100 / 200 / 500 | GPT-6 Sol, from Instant through Extra High | GPT-6 Pro, powered by GPT-6 Astra |
The three Pro prices are on About ChatGPT Pro tiers. The US Go price of $8 and the Plus price of $20 are on the ChatGPT Go announcement. On 12 October 2026 the US App Store price observed for monthly Go was still $8. If checkout shows euros, yen, or another local currency, use the pricing page. Do not put the US figure on the invoice.
Thinking levels
The Help Center splits Medium, High, Extra High, and Pro in one table. With GPT-6 selected, Instant, Medium, High, and Extra High run on GPT-6 Sol. The Pro level switches to GPT-6 Pro, which is Astra.
| Plan | Medium and High | Extra High | Pro |
|---|---|---|---|
| Free, Go | No | No | No |
| Plus | Yes | No | No |
| Pro, Business, Enterprise | Yes | Yes | Yes |
On Plus and Pro, Instant does not step up by itself when a question gets harder. Typing “think harder” or “think deeply” does not switch the level either. ChatGPT can still switch for safety. Free and Go use Think for harder questions. The Help Center says Think uses GPT-5.6 Luna, not GPT-5.6 Sol and not GPT-6 Sol.
Context and the other tools
These are the windows on the pricing page that day. The share you can fill is smaller than the total window, because system instructions, memory, and internal processing take space. Page counts are the pricing page’s approximations.
| Free | Go | Plus | Pro | |
|---|---|---|---|---|
| Instant total window | 27K | 54K | 54K | 128K |
| Instant input, about | 12 pages | 40 pages | 40 pages | 250 pages |
| Reasoning total window | Varies | 256K | 256K | 400K |
| Reasoning input, about | Varies | 320 pages | 320 pages | 680 pages |
Free text chat is not capped by message count. Uploads, image generation, voice, data analysis, deep research, memory, Codex, and ChatGPT Work on the desktop app each keep a separate limit. Go raises those limits and lengthens memory. The pricing page says Go may include ads. Plus adds Projects, scheduled tasks, custom GPTs, Work on web and mobile, and a wider Codex allowance. Pro adds three usage tiers, the always-on agent Dot, faster image generation that is not rationed by count, and a research preview.
The pricing page has a separate row, Interactive tables and charts: No on Free and Go, Yes on Plus and Pro. That row is not the same sentence as “Luna also supports Intelligent UI.” A free account can still show buttons and simple graphics. Treat interactive tables and charts as this row, then check what actually renders in the reply.
How Intelligent UI sits on the plan
Intelligent UI is a way of answering inside Chat, not a second subscription. GPT-6 composes text, graphics, tappable buttons, forms, charts, and a small tool you can use in the thread. A simple question can still come back as text only. Under it is a component library and a compiler, so the interface can appear while the model is still generating.
On questions that need web search, OpenAI’s own evaluation says GPT-6 Instant starts answering 44% sooner, on average, than GPT-5.6 Instant. That measures the start of the answer, not the finished reply. A second internal evaluation says that on high-value everyday agentic tasks, GPT-6 Extra High begins answering in the same time as GPT-5.6 Medium and scores higher than GPT-5.6 Extra High. Those are OpenAI’s internal numbers, not a third-party leaderboard, and they are not a Luna-versus-Sol accuracy gap.
The Help Center is explicit: GPT-6 Luna on Free and Go, and GPT-6 Sol on Plus, Pro, Business, and Enterprise, both support Intelligent UI. Astra on the Pro thinking level does not. If you want the interface, do not push the slider to Pro.
Where Intelligent UI stops
Surfaces that do not have it
- Chat only. The Work tab does not have it.
- Not in Voice. A voice call does not render this interface.
- Not on the Pro thinking level. Instant through Extra High have it. Pro is GPT-6 Astra.
- Not in the older desktop apps. The web app and updated supported mobile apps have it. The Help Center says the older macOS and Windows desktop apps do not, and tells you to use the web. Update the phone app to the current available version.
- Enterprise follows the admin. If the workspace has not enabled the model, it will not appear in a member’s menu.
- Still in the queue means absent. If GPT-6 is not in the menu, a different question will not produce the interface.
- Not an API. Public materials do not describe Intelligent UI as a callable endpoint. A calculator inside a thread is not an exported front end.
It does not fully turn off
The Help Center gives two web paths. Use the label on your screen. The Intelligent UI article says Personalization → Layout and Visuals → Simple. The model article says turn off Layout and visuals in Settings. Both say this only reduces visual and interactive elements. Some can still appear. It does not become a text-only mode. Custom Instructions can ask for short answers or for tables when comparing options. The layout can still vary. Inside one thread you can also say “text only.”
Quota and saved state
Intelligent UI has no separate quota. Existing model, tool, and plan allowances still apply. Making one thread less visual does not change your other threads.
Some components, such as checklists, can keep their state when you refresh that same thread. State is not kept across threads. Plugins you have connected, such as Gmail, Health, and Finance, keep working. A change in layout does not disconnect them.
Intelligent UI stops at Extra High. The Pro level in Chat is Astra, and the Help Center says that level does not support the interface. For buttons, charts, and in-thread tools, leave the slider at Extra High or below. For harder Astra work, accept an answer without Intelligent UI.
The Pro weekly cap
A Pro subscription is not unlimited GPT-6 Pro. The Help Center lists a Chat allowance on its own. Work and Codex are separate and do not draw from this table.
| Plan | GPT-6 Pro in Chat | How it shares with GPT-5.6 Sol Pro |
|---|---|---|
| Pro $200 | 200 messages per week | GPT-5.6 Sol Pro has a separate 170 messages per day. Together the two models are also capped at 200 messages per day |
| Pro $100 | 50 messages per week | Shared with GPT-5.6 Sol Pro. Switching between them does not add messages |
| Business Standard | 15 messages per month | Both models share those 15 |
| Business Premium | 50 messages per week | Both models share those 50 |
On Pro $200, once the weekly GPT-6 Pro allowance is used, Chat switches to GPT-5.6 Thinking at Medium. To keep using GPT-5.6 Sol Pro, select GPT-5.6 Sol in the model menu and then choose Pro, and only while that model’s daily allowance and the combined daily cap both have room left. Hitting the cap does not cancel the subscription. When a reset time is available, Chat shows it. Support does not reset the allowance.
On Plus, GPT-6 Astra is in ChatGPT Work and Codex, not on the Pro slider in Chat. Astra requires Codex CLI 0.153.0 or newer. If the desktop app does not show Astra, open Menu → Check for Updates, even if an automatic update just landed.
Ultrafast exists only on Pro $500, for GPT-6 Astra and GPT-6.1 Sol. It consumes included usage at 8× the Standard rate. After that allowance, credits are charged at 6× the Standard credit rate for the selected model. That is a consumption rate, not a speed guarantee. Fast consumes at 2.5×. Pro $100 and Pro $200 do not include Ultrafast.
An account that had an active Pro $200 subscription at any point from 22 September through 10:00 a.m. Pacific Time on 29 September 2026 can keep the previous included allowance through 29 October, while that subscription stays active. After that date the included allowance drops, and the price stays $200. The older allowance is whatever OpenAI emailed that account. A new Pro $200 subscription uses the table above.
Plus at $20 a month already includes Astra in Work and Codex, plus Sol at Medium and High in Chat. What it does not include is Extra High, and the Pro level in Chat. If those 50 messages a week will sit unused, do not jump from $20 to $100 for the word Pro.
Upgrade only for the level you use
- Stay on Free. Everyday questions, short text, rare uploads. Text chat is not capped by count. You do not get Sol at Medium or High. Uploads, images, voice, deep research, memory, and Codex stay limited. Work on the desktop app stays limited too.
- Look at Go, $8 a month in the US. You need more uploads, images, voice, and longer memory, and you still do not need to pick Medium or High yourself. This is not the GPT-6 Sol upgrade. The pricing page says the plan may include ads.
- Move to Plus, $20 a month. Writing, research, and tables; you want to choose Medium or High; you want Projects, scheduled tasks, and custom GPTs; you want Work on phone and web; you want a wider Codex allowance. Most people stop here.
- Move to Pro $100. The work actually stalls at Extra High, or you need a few dozen Astra-level Pro turns a week. The cap is 50 messages a week, shared with GPT-5.6 Sol Pro.
- Move to Pro $200. That Pro thinking is the daily job, and 50 messages are not enough. Chat includes 200 GPT-6 Pro messages a week. GPT-5.6 Sol Pro has its own daily allowance, and the two models together stay under 200 messages a day.
- Look at Pro $500 only when Ultrafast is the bottleneck. The 8× rate burns the included allowance faster. If you do not need that speed, $500 does not buy a new interface. The interface is already on the lower thinking levels.
The step where Codex still needs a Mac
If you only ask GPT-6 in the browser, the machine line is $0. Plus and Pro put GPT-6 Astra in Codex and ChatGPT Work. When the command has to be xcodebuild, the Simulator, or signing, a Linux sandbox cannot do it, and that step adds a Mac. On the day this was written the homepage starts at $101.7 per month for M4 · 16GB / 256GB and $200.3 per month for 24GB / 512GB. Checkout is the plan page. The starting price is not a contract price. The ChatGPT fee does not include the machine.
Keep the model, the tools, and the machine on three lines. How to split them is in the agent monthly cost article. Do not bill the same command to two machines.
FAQ
Can the free plan use GPT-6?
Yes, for everyday text chat. The 7 October product page says Free and Go move to GPT-6 Luna on 8 October, and the pricing page describes Free as unlimited GPT-6 text chats. The Help Center also says Instant and Think on those plans may still be GPT-5.6 Luna, and that they do not include Medium, High, Extra High, or Pro. Whether your menu says GPT-6 or GPT-5.6 Luna is what you see after you sign in. Neither one is GPT-6 Sol on Plus.
What is the difference between Plus and Pro?
Plus is $20 a month: GPT-6 Sol in Chat at Instant, Medium, and High, with no Extra High and no Pro level in Chat. Astra is in Work and Codex. Pro, from $100, adds Extra High and GPT-6 Pro (Astra) in Chat. Pro $100 is 50 messages a week, shared with GPT-5.6 Sol Pro. Pro $200 is 200 GPT-6 Pro messages a week, with a separate 170 GPT-5.6 Sol Pro messages a day, and a combined cap of 200 messages a day across both models.
Can Intelligent UI be turned off?
Not completely. On the web, use Personalization → Layout and Visuals → Simple, or turn off Layout and visuals in Settings, whichever label you see. The Help Center says this reduces visual and interactive elements, some can still appear, and it does not become a text-only mode. Inside a single thread you can ask for text only.
Why does Pro thinking drop the interactive interface?
The Pro thinking level runs on GPT-6 Astra. The Help Center says Intelligent UI covers Instant through Extra High and does not include Pro. For buttons, charts, and in-thread tools, leave the slider at Extra High or below. Voice and the Work tab do not have the interface either. The older macOS and Windows desktop apps do not have it; use the web.
Does upgrading replace a Mac with Codex?
No. Plus and Pro include GPT-6 Astra in Codex and Work. That edits code and runs tasks. xcodebuild, the Simulator, and signing still need macOS. That step adds a Cloud Mac. The 16GB tier starts at $101.7 a month on the homepage. If you only chat in the browser, do not add that line.
Match the menu, then pick the monthly fee
Free covers everyday text chat. It does not cover Sol at Medium and High. Plus at $20 covers Sol at Medium and High, plus Astra in Work and Codex. It does not cover Extra High or the Pro level in Chat. The Pro weekly cap is 50 or 200 messages, not unlimited. Intelligent UI is absent on the Pro level, in Voice, and in Work, and it does not fully turn off on the web. Add a Mac only for the command that cannot run once it leaves macOS.
The Chat fee does not include this Mac
GPT-6 in the browser does not need an always-on machine. When Codex reaches xcodebuild, the Simulator, or signing, put that step on a Cloud Mac that does not sleep. Keep the subscription and the rental on two lines.