Say the number out loud
The standard consumer plan for the main assistants has settled around $20 a month. Coding assistants sit lower, around $10 to $20. Business and team plans run roughly $25 to $60 per person.
Twenty dollars is about 1,700 rupees, 30,000 naira, 400 pesos, 320,000 rupiah. In much of the world that is a real decision, not a rounding error, and anyone who talks about it as trivially cheap is not talking to you. So let us be exact about when it earns its keep.
Four honest reasons to pay
You keep hitting caps. This is measurable. For two weeks, note every time you are blocked or downgraded mid-task. If it happens more than three or four times a week, you are paying in interrupted work already.
You need a specific paid-only capability. Larger file uploads. Running code. Connecting to your documents. A much longer context. Project workspaces that keep instructions across chats. Look at the feature you actually need and check whether it is behind the wall.
The time is worth more than the money. If a tool saves you three hours a month and three hours of your time is worth more than the fee, the arithmetic is finished. For a freelancer billing anywhere, this is usually a short conversation. For a student on no income it may genuinely not be.
You need better terms for your data. Business and team tiers generally come with a contract that your content is not used for training. If you handle client or patient information, that clause may be the entire reason to upgrade — not the model.
Four reasons not to
You use it twice a week. Free covers you. Nothing else to decide.
You are buying "the best model." The gap between the best paid model and a good free one is small on most everyday work and invisible on some of it. Run your five prompts from the last lesson before you assume otherwise.
You are paying for three at once. This is common and mostly waste. Pay for one, use two others on their free tiers as second opinions. You lose almost nothing.
The feature you want is free somewhere else. Image generation, transcription, translation and coding help all have capable free or cheap options from different companies. Paying one vendor for everything is convenient, not optimal.
The option most people skip
Almost every provider sells access by the token through an API, and you can use it from free open-source chat apps that run in your browser or on your desktop. You pay for what you use.
The rough shape of it: small and mid-sized models cost well under a dollar per million input tokens. Frontier models cost several dollars per million input tokens and more for output. A million tokens is roughly 750,000 words.
So a person who asks fifteen moderate questions a day, mostly on a mid-sized model, may spend $2 to $5 a month instead of $20. A person feeding whole codebases to a frontier model every day can pass $20 before the first week ends.
Two honest warnings. There is no cap unless you set one — go into the billing page and set a hard monthly limit on the first day. And you are now responsible for the product layer: no memory, no polished mobile app, no support, unless the client you chose provides it.
A test you can run this month
Keep a note for two weeks. Every time you are blocked by a limit, or you want a feature you do not have, write one line. At the end, count.
More than eight lines: pay for the plan. Two to eight: try the API route first. Under two: stay free, and spend the money on something else.
And cancel without ceremony when it stops earning its place. Your conversation history stays; you drop to the free tier; you can come back next month. There is no loyalty being tested here.
Before you move on