What are the ChatGPT Go limits?

Go has unlimited everyday text chats on GPT-6 Luna, a 54K-token context window on the default model and 256K on reasoning models. It gives more image creation and memory than Free, with deep research and Codex still limited, and at $8 a month it may show ads (OpenAI pricing page, October 2026).

Does ChatGPT Plus have a message limit?

Plus has no cap on everyday text chats, subject to OpenAI's abuse guardrails. Codex and ChatGPT Work run on rolling five-hour windows, which OpenAI estimates at 5 to 45 GPT-6 Astra messages per window depending on the task. Uploads, images and deep research have higher allowances than Go, with no published number.

Why does ChatGPT forget what I said earlier in a chat?

The conversation has outgrown the context window, the amount of text the model can hold at once. On the default model that is 27K tokens on Free and 54K on Go and Plus. Switching to a reasoning model raises it to 256K on Go and Plus, and starting a new chat with a short summary also works.

Can you turn off ads on ChatGPT Free?

Yes. Free users can switch to an ads-free mode that comes with lower usage limits, according to OpenAI's ads help article. Go subscribers have no ads-free option short of moving to Plus at $20 or dropping to Free.

When do ChatGPT limits reset?

It depends on the feature. Codex and ChatGPT Work on Plus reset on rolling five-hour windows, and Pro reasoning on the Pro tiers is reported to reset weekly. OpenAI publishes no reset schedule for uploads or image creation on Free.

ChatGPT message limits: what each plan caps

Everyday text chats are unlimited on every ChatGPT plan. Uploads, images, voice, deep research, reasoning and context are capped. The limits, by plan.

Is there a ChatGPT message limit?

ChatGPT has no fixed cap on everyday text chats on any plan, Free included. OpenAI's pricing page lists them as unlimited, subject to abuse guardrails. The limits sit on the extras: file uploads, image creation, voice, deep research, Codex, the strongest reasoning models and the context window. Those vary by plan, and OpenAI publishes exact numbers for only a few.

Many guides still quote caps such as 10 messages every five hours on Free or 160 every three hours on Plus. Those figures described earlier models. By late August 2026 OpenAI's pricing page listed text chats as unlimited on Free and Go, and as of October 2026 it lists the same on every paid plan.

ChatGPT limits by plan

	Free	Go	Plus	Pro
Price	$0	$8 a month	$20 a month	$100, $200 or $500 a month
Everyday text chats	Unlimited*	Unlimited*	Unlimited*	Unlimited*
Default model	GPT-6 Luna	GPT-6 Luna	GPT-6 Sol	GPT-6 Sol, plus Pro reasoning on GPT-6 Astra
Context window, default model	27K tokens	54K tokens	54K tokens	128K tokens
Context window, reasoning models	Varies	256K tokens	256K tokens	400K tokens
Image creation	Limited	Included	Included	Unlimited*
Deep research	Limited	Limited	Included	Maximum
Codex	Limited	Limited	Included	Expanded
Memory of past chats	Limited	Included	Expanded	Expanded
Response speed	Limited by bandwidth and availability	Limited by bandwidth and availability	Fast	Ultrafast on Pro 500
Ads	May show	May show	None	None

Source: OpenAI's pricing page, October 2026. *Unlimited subject to abuse guardrails, in OpenAI's wording.

The limits OpenAI publishes as numbers

Context windows are the one limit OpenAI states in figures for every plan. The context window is how much of the conversation, plus anything you paste or upload, the model can hold at once. OpenAI's rule of thumb is that a token is about three-quarters of an English word, so Go's 54K tokens comes to roughly 40,000 words and Free's 27K to roughly 20,000.

The Pro tiers have published terms too. OpenAI's DevDay recap put Pro 500 at 25 times Plus's usage, and its "About ChatGPT Pro tiers" help article (updated September 29, 2026) says new Pro 200 subscriptions carry a lower allowance than before. OpenAI's Codex pricing page states that "Pro plans currently have no five-hour limit." The full tier breakdown is in ChatGPT Plus vs Pro.

The same Codex pricing page estimates how many Codex messages Plus allows per five-hour window: 5 to 45 messages on GPT-6 Astra, 15 to 160 on GPT-6.1 Sol, 15 to 150 on GPT-6 Sol and 350 to 3,000 on GPT-6 Luna, depending on how large each task is. A standard ChatGPT Business seat gets a five-hour window too. Premium Business seats, at $100 a user a month billed annually, get 5 times a standard seat's usage and no five-hour limit, according to OpenAI's August 2026 announcement.

The limits that are reported, with no OpenAI number

These figures come from press coverage and OpenAI's emails to subscribers, not from a page OpenAI publishes. Treat them as approximate.

Pro-model chat messages on Pro 200: 200 a week, falling to 100 a week from October 30, 2026, according to OpenAI's email to existing subscribers as quoted by pricing trackers.
Usage multiples for Pro 100 and Pro 200 against Plus: 5 times and 20 times until late September 2026, with Pro 200 reported at 10 times for new subscribers since. For uploads, image creation and deep research on Free, Go and Plus, OpenAI gives words only (limited, more, expanded, maximum). It also gives no threshold for its abuse guardrails. Anyone quoting a precise daily figure for those is estimating.

What happens when you hit a ChatGPT limit

Text keeps working. Because everyday chat is unlimited on every plan, running out of a tool leaves you able to keep typing.

The tool itself pauses. When you reach the cap on uploads, image creation, deep research or Codex, ChatGPT shows a notice and the feature comes back when its allowance resets. Codex and ChatGPT Work on Plus reset on rolling five-hour windows. On paid plans, Codex and ChatGPT Work can keep going on purchased credits once the included usage is spent (OpenAI's Codex pricing page, October 2026).

Free and Go slow down at busy times. OpenAI describes their response times as limited by bandwidth and availability, so at peak hours answers can be slower or a feature can be temporarily unavailable.

Ads have a limit trade of their own. OpenAI's ads help article says Free users can switch to an ads-free mode with lower limits. Go subscribers have no such switch, so on Go the ads stay unless you move to Plus or drop to Free. Ads in AI apps covers how those ads work.

The limit with no counter: the context window

The context window gives no notice on screen. When a conversation grows past it, the model keeps answering but loses the earliest parts.

I have ChatGPT Go through my Revolut subscription, and I have hit this: partway through a task, the chat lost something I had told it earlier in the same conversation. I was on the default Instant model, where Go's window is 54K tokens.

Two fixes help. Switching to a reasoning model raises the window to 256K tokens on Go and Plus. Starting a fresh chat with a short summary of the facts and constraints that matter resets the count. What Go does and leaves out beyond this is on is ChatGPT Go worth it.

Choosing a plan around the limits

Pick by which limit you actually hit. If it is a tool (images, deep research, Codex), the next plan up raises that allowance, and whether ChatGPT Plus is worth it weighs the $20 step. If you hit the Codex or ChatGPT Work limits on Plus most weeks, Pro removes the five-hour window and raises the allowance. Other providers meter too: Claude Pro's usage limits run out for heavy users at $20.

If all you need is a lot of everyday chat, ChatGPT Free already gives you unlimited text, with ads and a 27K-token window. Plainly is another option for that job: $5 a month for unlimited chat with no ads, and long chats keep working. It does no image creation, voice or deep research, so it covers none of the tool limits above.

These figures are as of October 11, 2026. OpenAI changed its limits several times this year, so check chatgpt.com/pricing for the current version.