AI 101Beginner18 min read

Free vs paid: what you really get from an AI subscription

Why to start from the job, the dials that usually differ between free and paid, what a paid plan does not change, how to read a plan page, a seven-day test, a decision rule, and the money and safety traps around subscriptions.

Free vs paid: what you really get from an AI subscription

Free vs Paid: What You Really Get From an AI Subscription

Every AI tool now has a free plan and a paid one, and often several paid ones with names that change every few months. The marketing says "unlock the best model", "unlimited", "pro". It is hard to tell what you would actually get for the money, and easy to end up paying for three subscriptions that overlap. This tutorial does not list plans or prices, because those go out of date within weeks. It teaches something that lasts: what usually differs between free and paid, what usually does not, how to read a plan page without being fooled, a seven-day test that shows whether an upgrade would pay for itself, and a simple rule for deciding. You will finish with a one-page worksheet you can reuse every time a tool changes its plans.

Difficulty: ★☆☆☆☆ (no technical background needed)
Tools needed: Any AI tool you use or are thinking of paying for
Last updated: November 2026

Overview

Buying software used to be simple: you paid once, and you got the thing. AI subscriptions are different in a way that confuses people. The thing you are buying is not a fixed product. It is a bundle of access, and the bundle includes several separate dials.

One dial is how much you can use it: how many messages, how many long tasks, how many files, per hour or per day or per month. Another is which models you can reach, since a tool may offer a faster, lighter model and a slower, more capable one. A third is which features are switched on: deep research, image and video generation, voice, connections to your email and calendar, agents that act for you. A fourth is how your data is handled. A fifth is who it is for: an individual, a team or a company, with different administration and privacy controls.

A free plan usually gives you a taste of each dial. A paid plan turns one or more dials up. The key insight of this tutorial is that you should pay only for the dial you are actually running into. If the limit you hit is usage, you need more usage. If it is a feature, you need that feature. If you are not hitting a limit at all, a higher plan gives you nothing you will use.

Plans and prices in this area change quickly. Names are renamed, tiers are merged, features move from paid to free and back. So this tutorial gives no list of plans, no prices and no ranking of tools. If you read a table of plans in an article from last quarter, treat it as history. What follows is a method you can apply to whatever the pages say on the day you read them.

The tutorial has seven steps. First, why "free or paid?" is the wrong first question. Second, what usually differs. Third, what usually does not. Fourth, how to read a plan page. Fifth, a seven-day test. Sixth, how to decide. Seventh, how to protect yourself from the money and safety traps around subscriptions.

Who This Is Useful For

  • People using a free AI tool who wonder whether they are missing out.
  • People already paying for one or more AI subscriptions who suspect they are paying twice for the same thing.
  • Small business owners and freelancers deciding what belongs in their tool budget.
  • What You Will Learn

  • Why you should start from the job, not the plan.
  • The dials that usually differ between free and paid.
  • What a paid plan does not change.
  • How to read a pricing page, including "unlimited", trials and renewals.
  • How to run a seven-day test with a simple log.
  • A decision rule for upgrading, staying and cancelling.
  • What You Need

    1. At least one AI tool you use now, on a free plan or a paid one.
    2. A list of three to five real jobs you use AI for, such as drafting emails, summarising reports or researching a topic.
    3. A notes app or paper, for a seven-day log.
    4. Twenty minutes to start, then two minutes a day for a week.

    Information as of 14 November 2026. This tutorial deliberately names no plans, prices or limits, because they change often and differ by country. For current details, open the tool's own pricing page or the plan screen inside the app, and note the date you read it. If an article or video gives specific numbers, check them against the official page before relying on them.

    Step 1 — Start from the job, not the plan

    The most common mistake is to ask "should I upgrade?" before asking "what am I trying to do?". The first question has no answer. The second one does.

    Write down the jobs you use AI for, in a short list.

    JobHow oftenWhat matters most
    Drafting emailsDailySpeed and tone
    Summarising long reportsWeeklyHandling long documents accurately
    Researching a topic with sourcesA few times a weekCurrent information with links
    Making images for postsWeeklyQuality and the number of images
    Planning and brainstormingDailyMostly just a decent chat

    Now look at the right-hand column. Most everyday jobs, such as drafting, rewriting and brainstorming, work well on a free plan of a good tool. Upgrades matter most for jobs that are heavy (very long documents, lots of files), frequent (many tasks a day) or specialised (video, advanced research, agents).

    There is also a question that comes before plan and before brand: what kind of tool is this job for? A chat assistant, a search tool with sources, a tool that works from your own documents, an image generator and an agent that acts in your apps are different kinds of tools, and a plan for one does not give you the others. The next lessons in this module cover choosing the kind of tool. Start there, then look at plans.


    Step 2 — What usually differs between free and paid

    These are the dials, in rough order of how often they matter. Treat the table as a checklist of things to look for, not as a description of any one tool.

    DialWhat can differWhy it matters to you
    Usage limitsNumber of messages, long tasks, uploads or generations per hour, day or month; what happens when you run outThe most common reason people upgrade. Hitting a limit in the middle of work is the pain to look for
    Model accessWhich models you can choose; whether the most capable one is included, limited or paid-onlyAffects quality on hard tasks such as long reasoning, tricky code or careful writing
    FeaturesDeep research, image and video generation, voice, file and data analysis, connections to email and calendar, agents, longer outputsDecides whether the tool can do the job at all
    Context and memoryHow much text it can take at once, and whether it keeps notes or Projects between chatsMatters for long documents and ongoing work
    Speed and priorityFaster answers, or access at busy timesMatters if you use it constantly or at peak hours
    Data handlingWhether chats may be used to improve models, and the controls to turn that off; retention and deletionMatters for anything private or confidential
    Team and adminShared workspaces, user management, central billing, security settingsMatters if more than one person is involved
    Commercial terms and supportRights to use outputs commercially, support channelsMatters if the work is for clients

    Three things to keep in mind about the table.

    1. Each tool bundles these differently. A feature that is free in one tool may be the main reason to pay in another.
    2. Free plans often change. A free feature can become paid, or the other way round. A comparison from last season may be wrong.
    3. Data handling deserves its own look. Consumer plans and business plans commonly treat your data differently, including whether your conversations may be used to improve the product. Read the privacy settings and the policy on the official site, whichever plan you use, and do not paste confidential material into a tool whose terms you have not checked.

    Step 3 — What a paid plan does not change

    Knowing what money does not buy is as useful as knowing what it does.

    It does not make the AI truthful. A paid model can still invent facts, sources and quotes. It may do so less often, but the habits from the hallucination tutorial still apply. Paying is not a substitute for checking.

    It does not replace good prompts. A clear request with context and a format usually improves a free model more than an upgrade improves a vague request. Learning to prompt well is the cheapest upgrade there is, and the next module covers it.

    It does not remove the context window. A bigger plan may allow more text, but long chats still drift, and the habits for managing them still help.

    It does not guarantee privacy. A paid plan may give more control, but you still need to look at the settings and what you share.

    It does not give you a skill. An AI that can produce a video or a spreadsheet does not make the result good. You still need to know what a good result looks like.

    It does not make a tool right for the job. A premium chat assistant is still a chat assistant. If the job needs current facts with sources, your own documents or action in your apps, the right kind of tool matters more than the tier.

    One more thing that does not change is that the free plan of a good tool is genuinely useful. Plenty of people do excellent work on free plans. Paid plans are about limits and extras, not a different universe.

    Step 4 — How to read a plan page

    Pricing pages are written to make an upgrade feel natural. Read them like a checklist, with a pencil.

    1. Find what is measured. Messages? Credits? Hours? Tasks? If the page does not say clearly what a unit of use is, note that. Vague units make it hard to know what you are buying.

    2. Be careful with "unlimited". It can mean genuinely no cap, or "subject to fair use" with limits that are not stated. Look for a footnote or a link to usage policies. If a cap exists but is not stated, assume you may meet it.

    3. Check what happens at the limit. Do you wait until it resets, drop to a smaller model, or have to buy more? The consequence matters as much as the number.

    4. Look at the reset period. Limits that reset every few hours feel very different from limits that reset monthly.

    5. Check which models and features are included, and which are extra. Some features sit behind a higher tier, or cost additional credits even on a paid plan.

    6. Read the trial terms. How long is it, does it need a card, and does it renew automatically at the full price? Note the date it ends in your calendar the day you start.

    7. Compare monthly with annual. A yearly price may look cheaper per month but locks in your money, and a tool you stop using in month three has cost you the whole year.

    8. Check the currency, tax and region. Prices and availability vary by country, and what you see may include or exclude tax.

    9. Watch for app and API differences. A subscription to a chat app is usually separate from paying for developer access through an API. They are billed differently. Do not assume one covers the other.

    10. Note the date. Write "read on" and today's date next to anything you copy down. In three months it may be wrong.

    If the page is unclear after all this, that is itself information. A plan you cannot understand is hard to judge, and you can wait.

    Step 5 — The seven-day test

    The best evidence of whether an upgrade is worth it is your own use. Run a simple seven-day test on the plan you have.

    What to log. For each use, write one line:

    Date | Job | Tool and plan | Did I hit a limit? (yes/no, what happened) | Was the quality good enough? (yes/no, why) | Time saved (guess, in minutes)

    Keep it light. Two minutes a day is enough. You are looking for patterns, not precision.

    What to watch for.

  • Limit hits. Count how many times you were stopped, slowed or sent to a smaller model. One or two in a week is normal. Daily interruptions in the middle of paid work are a strong sign.
  • Quality gaps. Note any job where the answer was clearly not good enough, and whether the cause looked like the model, the missing feature or a weak prompt. Try an improved prompt before you blame the plan.
  • Missing features. Jot down any time you wished for a feature the tool did not give you, such as handling a long file or generating a certain kind of image.
  • Time saved. A rough estimate is fine. It lets you compare the benefit with the cost.
  • Optionally, a trial of the upgrade. If the plan offers a trial, run the same three jobs on it for a week, noting the same things. Set a calendar reminder to cancel before it renews, and cancel the same day if you decide against it.

    At the end of the week, count the limit hits, the quality gaps and the missing features, and sum the time saved. These numbers, rather than a feeling, decide what you do next.

    Step 6 — How to decide

    Use a short decision rule, and write it down so it is not decided in a good mood.

    Upgrade if all of these are true:

    1. You have a specific limit or feature that blocks real work. Not "it would be nice", but "I was stopped four times this week in the middle of client work".
    2. A better prompt or a different kind of tool would not fix it. You tried first.
    3. The value is clearly more than the cost. A simple comparison: if the plan costs a given amount a month, would the time it saves be worth more than that, using your own estimate of your time? If you cannot say yes quickly, the answer is probably no.
    4. You are willing to review it. Put a date in the calendar, one to three months ahead, to decide whether to keep it.

    Stay on free if:

  • You rarely hit a limit.
  • Your jobs are light: writing, rewriting, brainstorming, explaining.
  • The thing you want is a feature you would use once.
  • Cancel or downgrade if:

  • You have not hit a limit in a month or more.
  • Another tool you already pay for covers the same job.
  • You are using the plan for one feature you could get elsewhere more cheaply.
  • Your work changed and the original reason has gone.
  • Keep a "one paid tool at a time" rule. Start with one paid plan, for the dial that blocks you most. Add a second only when you can name a job it does that the first cannot. Most people do not need more than one or two.

    Check what you already have. Your employer, school, phone plan, internet provider or an existing software subscription may include access to AI tools. The cheapest plan is the one someone else is already paying for, as long as the terms allow your use and you are comfortable with who can see what.


    Step 7 — Money and safety traps

    A few traps are worth naming. None of this is financial advice, only common sense about subscriptions.

    Auto-renewal and free trials. A trial that asks for a card will usually charge when it ends. Set a reminder the day you start, and cancel before the end if you are unsure.

    Annual prepayment. Pay yearly only for a tool you have already used for several months and expect to keep.

    Stacking. Paying for two tools that do the same job is the most common waste. Before adding one, check what the other already does.

    Shared or "cheap" accounts from resellers. Offers of a low-priced shared account, or a plan bought through an unofficial seller, may break the provider's terms, may stop working without notice and put your account and data at risk. They are also a common route for scams. Buy from the official source.

    Sharing your own login. Check the terms before you share an account with family or colleagues. Many plans are for one person, and shared use can get an account suspended.

    Price and plan changes. Providers change plans and limits, sometimes without much notice. Keep your receipts, and look at your plan page occasionally to see whether what you pay for is still what you get.

    Privacy settings. Whatever plan you are on, look once at the data controls: whether conversations may be used to improve the product, how to turn that off and how to delete your history. A paid plan does not automatically switch on the setting you want.

    Work use. If you use AI for a job, check what your employer allows, and for client work, the commercial terms of the plan. Do not paste confidential material into a tool that is not approved for it.

    Regional availability. A plan or feature may not be offered in your country, and workarounds may break the terms. Check the official page for your region.

    Worked Example: Mei decides with a seven-day log

    Mei is a freelance writer. She uses a free plan of a chat assistant every day, and she keeps seeing ads for "the pro plan". She is tempted but unsure, so she runs the seven-day test.

    She lists her jobs: drafting article outlines, rewriting client copy, summarising research PDFs and brainstorming headlines. Then she keeps a one-line log for a week.

    By day seven she has this. She hit a usage limit twice, both times in the afternoon while summarising long PDFs, and was moved to a smaller model for a few hours. The quality of outlines, rewriting and headlines was fine on the free plan every time. The two PDF summaries were slower and, when she compared them, missed a point. A better prompt that asked for quotes improved both. She noted one wish: a way to upload several PDFs at once.

    She checks her options. The same one feature she wants, and the heavier usage, come with a paid plan. She works out that the PDF summarising saves her roughly three hours a week on weeks she has research projects, which is a few a month, and the plan costs less than an hour of her billing rate. She decides to upgrade, but only for three months, and she puts a review date in her calendar.

    She also cancels a second subscription she found in her audit, since it did the same thing as the first. At the review date, she looks at her log. She hit no limits during a quiet month. She downgrades, and plans to upgrade again only in heavy research months. Her total spend is lower than if she had signed up on day one and forgotten about it.

    Common Mistakes to Avoid

    Three habits cost people the most.

    Mistake #1: Upgrading to feel serious, not to fix a limit. "Pro" sounds like what professionals use, and the upgrade feels like progress. Fix: write down the specific limit or feature that blocks real work, try a better prompt or a different kind of tool first, and upgrade only for that dial.

    Mistake #2: Trusting a comparison article, or the plan page's headline. Plans change quickly, "unlimited" often has fine print, and an article from last quarter may be wrong. Fix: read the official plan page yourself, note the date, check what is measured and what happens at the limit, and treat any list from elsewhere as history.

    Mistake #3: Subscribing and forgetting. Overlapping subscriptions, forgotten trials and annual plans for tools you stopped using quietly add up. Fix: set a reminder when a trial starts, keep one paid tool at a time, and do a quarterly audit where every subscription must earn its place in one sentence.

    Going Further

    Make your own one-page plan worksheet. A table with the dials from Step 2 as rows, and the tools you use as columns. Fill it in from the official pages, with the date. Each quarter, update it in ten minutes.

    Measure time, not just feelings. For two weeks, time a few jobs with and without the paid features. A rough stopwatch beats a hunch.

    Look at free alternatives. For each paid feature you use, check whether another tool offers it free, or at a lower tier. Do not assume the first tool you tried is the best value.

    What to learn next. The next lesson in this module is about picking the right kind of tool for a job, a step that comes before choosing a plan. Then the quest on picking tools puts it into practice.

    Key Takeaways

    In this tutorial you learned:

  • Start from the job, not the plan. If you cannot name the job a paid plan would help with, do not buy it yet.
  • Plans are bundles of dials. Usage limits, model access, features, context and memory, speed, data handling, team and admin, and commercial terms.
  • Pay only for the dial you actually run into. A limit you hit, or a feature you need, not a feeling.
  • A paid plan does not make AI truthful, replace good prompts, remove the context window, guarantee privacy or choose the right kind of tool for you.
  • Read plan pages like a checklist. What is measured, the fine print behind "unlimited", what happens at the limit, trial and renewal terms, monthly vs annual, region and tax, app vs API, and the date you read it.
  • Run a seven-day test. Log limit hits, quality gaps, missing features and time saved, and let the numbers decide.
  • Decide with a rule. Upgrade for a specific blocker you could not fix another way and that is clearly worth the cost, with a review date. Stay on free for light jobs. Cancel what you do not use.
  • Avoid the traps. Auto-renewing trials, annual prepayment, stacking tools, unofficial resellers, shared logins and unchecked privacy settings.
  • Before you move on, a challenge. List your three most common AI jobs, and for one week keep the seven-day log on the plan you have now. At the end, write one sentence: "I will upgrade / stay / cancel because ___", and put a date in your calendar to look again. That single sentence, backed by your own numbers, is worth more than any comparison table.

    Learn AI, after work

    Track your progress, earn XP, and unlock more free tutorials in the AfterWork Academy app.

    Open this tutorial in the app

    More AI tutorials