Message limits
Gone for plain text since August 2026: everyday chat is unlimited on every plan, subject to abuse-prevention safeguards. What is metered instead is reasoning effort (Think, Medium, High, Extra High, Pro), by plan tier.
Independent guide · reviewed September 1, 2026
Hit a ChatGPT limit? Here is which one, and what to do.
Since August 2026, text chat is unlimited on every ChatGPT plan, including Free. The limits did not vanish; they moved to reasoning options, images, file uploads, voice, deep research and Codex. This guide maps each one per plan, with OpenAI’s own figures where they exist, and tells you what to do when you hit it.
Limit types
There is not one single ChatGPT limit.
People search for “ChatGPT daily limit” as if it is one number. In practice, ChatGPT can limit many separate things.
Model limits
You may have more access to standard models than to the most capable reasoning, research, voice or image models.
Tool limits
File uploads, image generation, voice, data analysis, connectors and deep research can each have separate quotas.
Context limits
Long files and long chats can exceed context capacity. The app may summarize, truncate or ask you to start a new chat.
File limits
Uploads can be limited by file size, file type, number of files and analysis time. Sensitive files also raise privacy questions.
Safety limits
Some requests are refused or restricted because of policy, not because your paid plan is too small.
Rate limits
Very fast repeated prompts, automation-like use or unusual traffic can trigger throttling.
Codex limits
Codex agent usage is different from normal chat: coding tasks can consume separate agentic usage and runtime resources.
Plans
Free, Go, Plus, Pro, Business and Enterprise: the map.
Official where OpenAI publishes a figure; “capped” where OpenAI confirms a limit but publishes no number. Prices as listed on chatgpt.com/pricing at review time.
| Plan | Price | Text chat | Reasoning | Images / files | Codex |
|---|---|---|---|---|---|
| Free | $0 | Unlimited (Luna) | Think only, limited | Capped; 3 uploads per day | — |
| Go | $8/mo, select countries | Unlimited (Luna) | Think only, limited | Higher than Free | — |
| Plus | $20/mo | Unlimited (Sol) | Instant, Medium, High | Higher; 20 files per project | Included, token-metered |
| Pro | $100 (5x) or $200 (20x) | Unlimited | Instant through Pro (Sol Pro) | Highest; 40 files per project | Included, most headroom |
| Business | Per seat | Standard seat per 5 h: Sol 10–100, Terra 25–200, Luna 250–2,000 (official estimates) | Instant through Pro | 40 files per project | Included; workspace credits |
| Enterprise / Edu | Contract | Per contract; Premium seats 5x, no 5-hour limit | Instant through Pro | Per contract | Included; credits |
Every plan shares the same hard file limits: 512 MB per file, 2 million tokens per document, about 50 MB per spreadsheet, 20 MB per image, 80 uploads per three hours. Details on the file upload page; reasoning tiers on the Thinking limit page.
Which limit?
Which ChatGPT limit do you actually mean?
The same words cover very different caps. Match the phrase to the allowance and the fix.
| What you searched for | What it means in 2026 | What to do |
|---|---|---|
| “ChatGPT limit reached” | A feature allowance, not a chat cap: text is unlimited. The message names the feature. | Read the message; keep working in plain chat; wait for the reset shown. |
| “ChatGPT free limit” | On Free, the Think option, images, voice, deep research or the 3-uploads-a-day cap. | Paste instead of upload; use plain chat; Go or Plus if it happens daily. |
| “ChatGPT daily limit” | There is no daily message limit any more. Feature allowances reset on their own windows (uploads: 80 per 3 hours; research: 30 days). | Trust the reset time in the app, not the clock. |
| “ChatGPT Plus limit” | The Medium or High reasoning allowance, or images, uploads, research or Codex. | Chat continues on a lower setting; escalate reasoning only for the hard step. |
| “ChatGPT Pro limit” | Reasoning up to Sol Pro, or Codex tokens, at 5x or 20x Plus headroom. | Scope Codex runs; use Instant for routine text. |
| “Codex limit” | Token-metered agent usage under OpenAI’s rate card, separate from chat. | Smaller, named tasks with a test command; see the Codex page. |
| “ChatGPT restricted” | A policy refusal or a workspace restriction, not usage. | Rephrase, or ask your admin; waiting changes nothing. |
When capped
What happens when you hit a ChatGPT limit?
You usually do not lose your work. You may need to wait, switch model, reduce the task, upgrade or move the work to another AI tool.
Wait for reset
If ChatGPT shows a reset time or quota counter, the cleanest move is to wait. Some limits reset after a rolling window, not exactly at midnight.
Switch model or tool
The app may offer a lighter model, fewer tools or a different mode. That can be enough for simple work.
Reduce the task
Shorten the file, split the prompt, remove unnecessary context or ask for an outline first. Long prompts burn scarce usage faster.
Upgrade legitimately
If the limit blocks normal work, compare Free, Go, Plus, Pro, Business and Enterprise on the official pricing page.
Use alternatives
Claude, Gemini, Perplexity, Copilot and all-in-one tools can help when ChatGPT is temporarily capped.
Do not evade rules
Creating extra accounts only to bypass limits can create account and terms risk. Use official options instead.
Real situations
The right move depends on what you were doing.
A message cap, an image cap, a file cap and a coding-agent cap are different problems. Treating them the same wastes time.
You are writing an email or short text
Do not waste advanced usage on tiny rewrites. Use a lighter model when offered, ask for one polished version, then edit manually.
You are uploading a long PDF
Ask for an outline first. Then ask questions section by section. Very large files can burn limits faster because extraction, context and analysis all matter.
You are generating images
Image limits can differ from chat limits. If image generation stops, text chat may still work, and the image quota may reset separately.
You are doing deep research
Research-style tools can have separate caps because they search, browse, synthesize and cite. For source-heavy work, Perplexity or other research tools may be more efficient.
You are coding with Codex
Keep tasks scoped. Ask Codex to inspect, patch and test one issue at a time instead of one huge vague request that consumes long agent runtime.
You are in a company workspace
Your admin can restrict connectors, uploads, sharing, retention, models or billing. The account limit may not be the only limit you feel.
Myths
Bad advice about ChatGPT limits causes account risk.
The safest answer is boring but useful: read the in-product message, use official plans, reduce heavy tasks, and avoid tricks that create terms or billing problems.
“A paid plan means unlimited ChatGPT.”
No. Paid plans usually raise access, but scarce models, advanced tools, safety rules and abuse-prevention systems can still apply.
“Daily limit means midnight reset.”
Not always. Some limits can behave like rolling windows, feature-specific counters or temporary capacity restrictions.
“A refusal means I hit the message cap.”
Not necessarily. A refusal can be about policy, missing context, privacy risk, unsafe instructions or restricted content.
“A VPN fixes limits.”
Do not build around that. Region, billing, policy and account rules can create new problems.
“More prompts always solve it.”
Often the opposite. Repeating the same large prompt burns more usage. Summarize, narrow the task and ask for one specific output.
“The API subscription is the same as ChatGPT.”
No. API usage, rate limits, token billing and ChatGPT web/app subscriptions are separate products.
Alternatives
Free alternatives and all-in-one AI tools.
If a cap lands mid-task and you cannot wait for the reset, the practical fix is a second tool. None of these are affiliated with OpenAI; pick whichever fits the job.
Multi-model
MultipleChat AI
multiplechat.aiBest when you want ChatGPT, Claude, Gemini, Grok and Perplexity-style search in one workflow instead of waiting on a single model limit.
AlternativeClaude
claude.aiStrong for long text, careful writing and document-heavy work.
AlternativeGemini
gemini.google.comUseful when work lives in Google products or Google search context matters.
ResearchPerplexity
perplexity.aiGood when the job is answer-with-sources research rather than open-ended chatting.
WorkMicrosoft Copilot
copilot.microsoft.comUseful for Microsoft users and Office workflows.
Multi-modelPoe
poe.comMulti-bot interface with access to several AI systems.
Disclosure: MultipleChat is built by the team that runs this site. It is listed as one option among several; nothing on this page is paid placement.
Codex
Codex limits are not the same as normal ChatGPT messages.
Codex is for software engineering agent work. It can read code, edit files, run commands and work across tasks. That means its usage is measured differently from a normal chat conversation.
| Product | What it is | Limit note |
|---|---|---|
| Normal ChatGPT | Conversation, writing, explanations, brainstorming, images, files, research and general assistant tasks. | Usually limited by messages, models, tools and plan-level access. |
| Codex | Coding agent work: reading repositories, editing files, running commands, tests and producing patches or PR-style changes. | Can have separate agent/task/runtime limits and may be tied to coding-focused plan access. |
| Codex CLI / app / IDE | Developer workflow surfaces that can connect to local or remote code workflows. | Not the same as asking ChatGPT a coding question in a chat box. |
| Enterprise Codex | Team or company coding automation with security, admin and procurement needs. | Expect organization-specific controls and limits. Confirm with OpenAI. |
Prompt fixes
Use smaller prompts before you burn more usage.
When you are close to a cap, the goal is not to send more. The goal is to preserve context, narrow the work and make the next model call count.
Shorten a task before retrying
“Summarize this request into the smallest prompt that preserves the goal, inputs, constraints and desired output format.”
Split a document job
“Create a section-by-section work plan for analyzing this document without needing to process everything in one response.”
Use a lighter model safely
“Give me a quick first draft only. Do not over-reason. Keep it concise and leave placeholders where facts need verification.”
Preserve work before a cap
“Before we continue, summarize all decisions, open questions and next steps so I can resume this in a new chat if needed.”
Move to another model
“Rewrite this prompt so I can paste it into Claude, Gemini, Perplexity or another assistant without losing context.”
Prepare for Codex
“Turn this coding problem into a narrow implementation task with files to inspect, expected behavior and test commands.”
Business use
Company limits are about more than message counts.
For teams, the important questions are often admin control, data handling, file permissions, connectors, procurement and whether users should rely on personal accounts at all.
Personal use
Free, Go, Plus or Pro may be enough. The main decision is whether you need more access, better models, file tools, voice, images or coding help.
Small team
Business or Team-style workspaces matter when billing, ownership, shared usage, connectors and admin controls become important.
Enterprise
Enterprise decisions are less about a public message number and more about procurement, security review, data controls, support, access management and contract terms.
Developers
Codex and API workflows should be planned separately from ChatGPT chat. Coding-agent runtime and API rate limits are different kinds of limits.
Restrictions
Is ChatGPT restricted? What about 18+ content?
Yes, ChatGPT has both usage limits and safety restrictions. Paying more can raise usage access, but it does not remove policy rules.
Restricted does not always mean “limit reached”
If ChatGPT refuses a request, the reason may be policy, safety or missing context rather than daily quota. A refusal is different from a usage cap.
18+ content depends on the request
Adult topics, health, relationships, education, explicit sexual content, minors, exploitation and non-consent are treated differently. Do not rely on this page as permission; check OpenAI policy.
New accounts are not a clean workaround
Creating extra accounts to bypass limits can create account-risk and terms-risk. It is safer to wait, upgrade, use allowed alternatives or reduce the task.
Companies need admin rules
Business and Enterprise users may face workspace restrictions, connector permissions, data controls, admin policy and procurement requirements beyond public ChatGPT plan limits.
Sources
Where to verify current limits.
Because limits change, this page links to official sources and relevant comparison pages.
Official ChatGPT pricing
https://openai.com/chatgpt/pricing/Plan names, plan positioning and current commercial wording can change.
OpenAI usage policies
https://openai.com/policies/usage-policies/Rules for prohibited content, sexual content, abuse, safety and policy enforcement.
OpenAI terms
https://openai.com/policies/terms-of-use/Account rules, service rules and user responsibilities.
ChatGPT app
https://chatgpt.comFinal in-product limits, counters, upgrade prompts and account state.
Codex in ChatGPT
https://chatgpt.com/codex/Coding-agent entry point when available on your plan.
Deep dives
Specific ChatGPT limit pages.
The main page explains the whole topic. These pages answer the searches people actually type after they hit a cap.
ChatGPT Free Limit Explained
What ChatGPT free limits mean, why free users hit caps, what resets, what lighter models do and what to use instead.
ChatGPT Daily Limit and Reset Time
How ChatGPT daily limits and reset windows work in practice, including rolling windows, feature-specific limits and why users see different caps.
ChatGPT Plus Limit Explained
ChatGPT Plus limits explained: why Plus is higher than Free but not unlimited for every model, tool, file, image or research task.
ChatGPT Pro Limit Explained
ChatGPT Pro limits explained for heavy users, reasoning models, Codex, research, images and abuse guardrails.
ChatGPT Business and Enterprise Limits
Business, Enterprise and Edu ChatGPT limits explained: credits, admin controls, flexible pricing, workspaces, Codex seats and governance.
Codex Limit and ChatGPT Coding Limits
Codex limits explained: agentic coding usage, token-based rate cards, credits and why Codex is not the same as normal ChatGPT messages.
ChatGPT File Upload Limit
ChatGPT file upload limits explained: file size, file count, PDFs, spreadsheets, images, context and privacy.
ChatGPT Image Generation Limit
ChatGPT image generation limits explained: why image caps differ from chat caps and how to reduce wasted generations.
What to Use When ChatGPT Limit Is Reached
Legitimate alternatives when ChatGPT limits are reached: Claude, Gemini, Perplexity, Copilot, Poe, MultipleChat and source-focused tools.
FAQ
ChatGPT limit questions answered.
Short, careful answers for people searching free limits, daily limits, plan limits, Codex limits and policy limits.
How many ChatGPT messages per day do I get?
Since August 2026, everyday text chat is unlimited on every plan including Free, subject to abuse-prevention safeguards. What is limited is reasoning (Think, Medium, High, Extra High, Pro depending on plan), images, file uploads (3 per day on Free, 80 per 3 hours on all plans), voice, deep research and Codex. The in-app message shows which allowance you hit and when it resets.
What happens when I hit the ChatGPT daily limit?
Usually you see a limit message, a wait time, an upgrade prompt or an option to continue with a lighter model. For some tools, the quota may reset on a rolling schedule.
Can I create a new account and continue?
That is not a good recommendation. Creating extra accounts to evade limits can violate platform rules or trigger account review. Use official limits, wait, reduce the task, upgrade or use legitimate alternatives.
Is ChatGPT restricted?
Yes. Some restrictions are usage limits, and some are safety or policy restrictions. A paid plan can raise usage access, but it does not remove safety rules.
Does ChatGPT allow 18+ content?
OpenAI policy is the final source. In general, adult topics may be handled differently depending on whether the request is educational, health-related, explicit, exploitative, illegal, non-consensual or involves minors. Do not treat any third-party guide as permission.
What is a ChatGPT free limit?
Free limits are lower and more variable than paid plans. Free users can hit caps for messages, advanced models, file tools, images, deep research or voice sooner than paid users.
What is a ChatGPT Go limit?
Go is a lower-cost paid tier in places where OpenAI offers it. It usually sits above Free and below Plus. Check the official pricing page for availability and current features in your region.
What are ChatGPT Plus limits?
Plus generally gives higher access than Free, but not unlimited access to every model and feature. Heavy users can still hit model, image, research, voice or file limits.
What are ChatGPT Pro limits?
Pro is intended for heavier personal use and usually has the highest individual access. Still, it can have fair-use, safety and resource guardrails.
What are ChatGPT Enterprise limits?
Enterprise limits are often negotiated or workspace-specific. Ask OpenAI or the account owner, and check the contract/admin documentation.
Are Codex limits the same as ChatGPT limits?
No. Codex is an agentic coding workflow, not just a chat. It can consume task/runtime/agent usage separately from normal messages.
Does Codex have daily limits?
Codex limits depend on plan and product surface. The safest wording is to say Codex has separate usage rules and to check the in-product limit messages or OpenAI plan page.
Can ChatGPT read unlimited PDFs?
No. File upload, size, file count, extraction quality, OCR, context length and analysis time can all limit document work.
Why did ChatGPT switch to another model?
When a quota is reached or demand is high, ChatGPT may offer or automatically use a lighter model. This is normal limit behavior.
Why does my friend have different limits?
Limits can differ by plan, region, rollout group, model, account history, workspace settings and current system demand.
Can a VPN change ChatGPT limits?
Do not rely on location tricks. Region availability, billing, policy and account rules can apply. Use official access paths.
Do API limits equal ChatGPT limits?
No. ChatGPT web/app subscriptions are different from OpenAI API rate limits, token limits and billing.
What should I use when ChatGPT free limit is reached?
For one-off work, wait or use a lighter model. For ongoing work, compare Plus/Pro, or use alternatives like Claude, Gemini, Perplexity, Copilot or MultipleChat.
Is MultipleChat a ChatGPT replacement?
It is better described as an all-in-one interface: you can work with multiple AI models in one place, compare answers and continue when one model is not the right fit.
Why does ChatGPT say try again later?
That can mean temporary capacity, a tool quota, a model limit, rate limiting, connection trouble or a safety review. Read the exact message before assuming it is a daily message cap.
Can I keep working for free after hitting a ChatGPT limit?
Sometimes yes, but not by evasion. You may be able to use a lighter model, wait for reset, use another AI with a free tier, or use another legitimate AI tool while you wait.
Do file uploads count differently?
They can. A short chat message and a large spreadsheet, PDF, image or code archive do not place the same load on the system. Uploads, extraction and analysis can have their own limits.
Do ChatGPT limits apply on mobile too?
The same account and plan rules generally matter across web and mobile, but app version, region, rollout and feature availability can affect what you see.
Can my employer see or control my limits?
In a workspace, company admins may control access, billing, connectors, retention, sharing and allowed features. Ask your admin before uploading company data.
Should I pay just because I hit one limit?
Not always. If it happened once, wait or reduce the task. If a cap blocks normal daily work, compare official plans and alternatives by workflow, not by one angry moment.
Who has the final word on limits?
ChatGPT.com, OpenAI’s pricing page, OpenAI’s policies and your account/workspace admin have the final word. This page is informational and may be wrong or outdated.