What limits apply to ChatGPT image generation, and how do I check my current allowance?
The free tier can generate images, but the image tool has its own separate usage limits. Judge current availability from an actual generation request and the message shown in your account, and distinguish upload limits, generation caps, and general failures instead of inferring image allowance from whether chat still works.
The ChatGPT free tier can generate images, but image generation has its own separate usage limits: chat still working does not mean images will. To judge whether it is currently available, check the message your account shows after an actual generation request; do not treat the daily image counts in older guides as your fixed allowance.
Where do you start generating images on the free tier?
The OpenAI free tier FAQ explicitly lists image generation as a feature. First sign in to the account you plan to use, submit a clear generation request in chat, and check whether what comes back is an image, a limit notice, or some other error.
Suppose you want a reading nook illustration. You could enter: "Please generate a reading nook illustration with a bookshelf, an armchair, and a floor lamp." This is only a sample test request and does not represent an actual generation result.
When checking availability for the first time, start with a text description and hold off on uploading a reference image, so the upload step does not complicate diagnosis.
How do you tell whether you still have allowance?
Check in this order:
Make sure you are actually requesting image generation. Asking it only to write an image prompt does not verify whether image generation is available.
Read the exact wording of the message after the request. OpenAI states that ChatGPT notifies users when they reach the applicable limit.
Record the limited feature and the time of the notice. If the page gives a recovery time, plan subsequent tasks around that notice; if there is no specific figure, do not work out a remaining image count yourself.
A successful generation confirms that this one request worked, but you cannot infer from it how many more images you can generate. Whether chat works normally is also no substitute for judging the image tool's allowance.
If uploading a reference image fails, does that count as the generation allowance being used up?
Do not merge the two into a single judgment. OpenAI lists "file and image uploads" and "image generation" separately as tools that can be limited independently.
If the failure happens while selecting or uploading a reference image, record the upload notice first; if the reference image has been submitted and the limit is only raised at the generation stage, record the generation notice. If you only get a "general failure" with no mention of a limit, keep the exact error text and do not rush to conclude that the allowance is exhausted.
This distinction prevents an upload restriction from being mistaken for all image features being unavailable.
After reaching the limit, how long is the wait?
The free tier documentation states that after a tool reaches its usage limit, you need to wait a while before using it again. Do not assume it resets at midnight every day, or that it recovers after a fixed number of hours.
If the notice states a time, plan around the time given for that feature; if it only says the limit has been reached, keep the notice and try again later. Do not apply reset times shown for other features to image generation.
Would upgrading or switching to the developer API solve it?
The free tier documentation states that when you upgrade from the free tier to a designated paid plan such as Plus, usage allowance already used up is reset. This does not mean image generation is unlimited afterward. The official Plus description lists image generation, and also notes that usage limits may change with system conditions.
The API (application programming interface, which lets programs call the service) is billed separately from a chat subscription; Plus does not include API usage, so a subscription cannot be treated as a balance for the image API.
If your image needs are often interrupted by limits, first record how often you need them and the exact notices you receive, then compare plans using the subscription overview; do not assume that "includes image generation" is enough for any number of tasks.
FAQ
Can the free tier generate images?
Yes. OpenAI's free tier documentation lists image generation, but the image tool has its own separate usage limits; chat still working does not mean images will.
If chat still works, does that mean I still have image allowance?
Not necessarily. File and image uploads and image generation are listed separately as tools that can be limited independently. Judge by the notice after an actual generation request; you cannot infer it from the state of chat.
If uploading a reference image fails, is the generation allowance used up?
You cannot conclude that directly. Look at which step failed: if the error occurs while selecting or uploading the reference image, record the upload notice; only if the reference image has been submitted and the limit is raised at the generation stage should you record the generation notice.
How long does it take to recover after reaching the limit?
Do not assume it resets at midnight or recovers after a fixed number of hours. If the notice states a time, plan around that time; if it only says the limit has been reached, keep the notice and try again later.
Does upgrading restore normal use immediately?
The official documentation states that when you upgrade from the free tier to a designated paid plan such as Plus, used-up allowance is reset, but this does not mean image generation is unlimited afterward.
Where you check your ChatGPT renewal date depends on where you subscribed: web subscriptions in your account settings, Apple in-app purchases in the system subscription list, Google Play on the Play Store subscriptions page, and self-serve Business in the workspace billing page. This article distinguishes the next charge date from the expiration date, gives the lookup steps for each channel, the 24-hour cancellation cutoff, and the order to troubleshoot in when you cannot find the date.
According to the OpenAI official supported countries list verified on 2026-11-20, Hong Kong is not included. The official rule is that the API is unavailable in locations outside the list, and access from outside the list may result in an account being banned or suspended. This article shows Hong Kong users how to check eligibility in three layers — official list, network exit, account status — and explains the order in which to handle a regional error and why a proxy is not a solution.
ChatGPT Business is OpenAI's self-serve ChatGPT workspace plan for teams, priced per seat: a minimum of 2 seats, with Standard seats at 25 USD per person per month on monthly billing or 20 USD per person per month on annual billing, and Premium seats at 125 USD per person per month on monthly billing or 100 USD per person per month on annual billing; API usage and credits beyond the included allowance are billed separately.
Gemini's cost is a subscription fee charged through Google One storage membership and Google AI plans. Every account starts with up to 15GB of free storage, paid tiers start at 100GB, and Google AI plans with AI features start at 200GB. Tiers are based on storage capacity and billed monthly or annually, and the final amount is determined by the tier, the billing cycle, and the billing location. This article checks the storage and availability of each tier and the steps to upgrade, change, or cancel. It also explains the ChatGPT top-up path that readers in Hong Kong and Taiwan can complete with Alipay or WeChat Pay.
The ChatGPT registration region restriction is a rule under which OpenAI only allows users to register accounts and create API keys in countries or regions on its official supported list. This article explains what error you see when registering outside the list, how OpenAI determines the registration region, and why phone number or identity verification must be completed by the account holder. It also sets out the boundary of our top-up service: we handle payment only, not registration or verification.
ChatGPT subscription fees are non-refundable by default. Only a few exceptions may be approved: billing errors confirmed by OpenAI, verified unauthorized charges, documented service or access failures, and an unused individual subscription within 7 days that meets the conditions. The EU, UK, Turkey, South Korea, and Brazil also have statutory refund rights. Accidental purchases, forgetting to cancel, dissatisfaction, and switching plans are not, in themselves, grounds for a refund. This article sets out, by channel and region, the situations in which you can and cannot apply.
Netflix bills monthly only, with no annual plan. Based on the Hong Kong page verified on 2026-11-18, the three ad-free tiers cost HK$78 (Basic), HK$98 (Standard) and HK$118 (Premium) per month, and each extra member costs HK$31 per month. If you only want 4K for yourself, the official service sells you a whole household with 4 slots, whereas our Hong Kong Premium 4K shared slot sells a single position for ¥33 per month and accepts Alipay and WeChat Pay, which is better suited to one person's own use.
MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external tools and data sources; it solves the problem of manually copying and pasting data from tools into the AI. This article explains what it is for, gives the four commands for adding, checking, using, and removing a server in Claude Code, a comparison table of the four connection methods, and the troubleshooting order when a connection fails. It closes with how readers in Hong Kong and Taiwan can complete the payment step for AI tool subscriptions using Alipay, WeChat, and RMB.
An AI Agent is an AI program that can break down a task itself, call tools and work through multiple steps until the job is done, which is quite different from a chatbot that only handles one question and one answer. This article explains the difference and a three-step way to tell them apart, gives the steps for anyone to start using their first agent, and explains how readers in Hong Kong and Taiwan can subscribe to ChatGPT with Alipay or WeChat when they have no usable foreign-currency credit card.
There is no fixed completion time for a top-up: account status, official risk control, queue conditions, and payment confirmation together determine how fast it goes. We do not make a fixed promise; the status on the order page is authoritative. This article breaks down the four factors and explains when to keep waiting and when to contact support.
A GPT version number is a model name; it is not the same as the list your account can actually call. Which models you can use is decided jointly by your plan tier, whether the model has been retired, and the allowance available at the time: older models such as 4o are being retired, Go does not include older models, Think uses GPT-5.6 Luna, and only Pro 500 includes Ultrafast. Always verify against the model picker.
According to two official Google documents checked on 2026-11-18, the Gemini pages list no student discount or student plan. What can be confirmed about family arrangements is Google One family sharing: storage, benefits and support shared with up to 5 family members, excluding the 30 GB Lite tier, and whether AI benefits extend to the family group must be confirmed item by item. This article answers whether there is any basis for a student price, what family sharing actually shares, why pricing can only be seen at checkout, and how region and age restrictions apply, with a tier comparison table, a five-step check order and alternative routes.