Which ChatGPT plan, honestly

Plan advice ages badly, so this page avoids quoting prices and message counts. They move, and a figure without a date is worse than no figure. What does not move is the shape of the decision.

The shape of the ladder

Individual tiers run Free, Go, Plus, then Pro. Pro itself has more than one level. On the organisational side there is Business, then Enterprise and Edu, which are quoted rather than listed. The API is billed by usage and is a different kind of thing entirely.

The useful detail is how allowance is expressed: messages within a rolling five-hour period, varying by which model you are using. That structure is what you will actually feel, more than the monthly price. Your experience of a plan depends on how bursty your work is, and two people paying the same amount can have completely different days.

The question that decides it

How often do you hit the window, and when?

Never: the tier you are on is doing its job, and the next one up buys you nothing you will notice.

Once or twice a week, usually late in a working session: one step up is the honest answer.

Daily, and early: you are working in a way the five-hour window was not sized for. That is the case where a larger multiple genuinely changes the day rather than postponing the same wall.

Where upgrading does not help

When the model is the constraint. If the answers are not good enough, more of the same model does not fix it. That is a model-selection question wearing a plan question's clothes.

When the constraint is a habit. Very long threads and unbounded agentic runs will consume whatever you give them. A bigger plan used the same way runs out later in the afternoon, not never.

When what you actually needed was the business tier's terms. If the reason you are considering an upgrade is that you keep pasting work material into a personal account, the fix is the account type rather than the allowance. Business and Enterprise workspaces carry the no-training-by-default commitment and admin-controlled retention; Free, Go, Plus and Pro do not.

What goes wrong

Buying the top tier to be safe. A month one step down, with attention paid to when you hit the window, tells you what you actually need.

Comparing plans by price alone. The five-hour window means the same price buys a different experience depending on how you work.

Treating an upgrade as a privacy fix. It is not one. The training and retention defaults change at the workspace boundary; every consumer tier shares them.

Reading a message-count table from a blog. Those are inferred and undated.

How to check it worked

A month after changing tier, ask whether you still hit a limit at the same point in the week. If the answer is yes, the plan was never the constraint, and the money is buying you a later wall rather than no wall. If the answer is that you have stopped noticing limits entirely, consider whether you have overshot and could step back down.