What does ChatGPT Plus actually give you over the free plan?

Here is the honest part most articles skip: OpenAI does not publish a simple side by side of Free versus Plus anywhere our collector can read it. Its main pricing page refuses automated readers and its help centre articles do the same. What OpenAI does publish, on its documentation pricing page, is the plan prices and a table of per model usage ranges for the paid tiers. So this guide tells you exactly what is published, what it means, and what OpenAI leaves unstated, rather than filling the gaps with guesses.

The prices, as published. Free is $0. Go is $8 a month. Plus is $20 a month. Pro starts at $100 a month. No annual prices are published for the individual plans.

The clearest published difference is the model table. OpenAI’s docs list GPT-6 Astra, its newest flagship, announced on 3 September 2026, with a Plus allowance of 5 to 45 local messages per five hour period. The Free plan does not appear in that table at all; for free users the page commits to nothing beyond suitability for quick tasks. If having the best available model, with a stated allowance, is why you would pay, that is the concrete thing $20 buys.

What “5 to 45 messages per five hours” really means

Two pieces of jargon are hiding in that phrase, so let us take them apart.

A model is the engine answering you. OpenAI runs several at once: the bigger ones reason better and cost more to run, the smaller ones are faster and cheaper.

The Plus tier lists seven models, from GPT-6 Astra at 5 to 45 messages down to GPT-5.4 mini at 60 to 350 messages. The pattern is a trade: the smarter the model, the fewer turns you get.

The range exists because OpenAI meters work, not turns. In its own wording, the number of messages you can send depends on the model used and the size and complexity of your tasks. Ask short questions and you live near the top of the range; paste in long documents and you slide toward the bottom. OpenAI adds that weekly limits may also apply, and that tasks run in the cloud can draw more from your allowance than local ones.

One caveat we insist on. OpenAI publishes these ranges on its developer facing docs page, in the context of coding agent usage drawn from your plan’s overall allowance. It is the only per plan, per model table OpenAI publishes anywhere we can verify. Treat the ranges as the best available official signal of how the tiers compare, not as a guarantee printed on the chat window.

So who should pay the $20?

Stay on Free if ChatGPT is an occasional question answering tool for you. The free tier’s limits are unpublished but real; if you meet them once a month, they are not worth $240 a year to remove.

Consider Go at $8 if you meet free tier limits regularly but do not care about having the newest model. Its published commitment is thin, a price and no allowance table, so you are trusting OpenAI on the detail.

Plus at $20 is for people who use ChatGPT most days for work or study, want GPT-6 Astra rather than an older model, and want stated allowances. Across a month that is about 65 cents a day.

Pro from $100 exists for heavy, sustained use: its tiers multiply the Plus ranges several times over. If you have never met a Plus limit, it is not for you.

What OpenAI does not publish

We track this page over time, so it is worth saying plainly what is missing as of 6 September 2026: no per model allowances for Free or Go, no plain chat message caps for any tier, no annual pricing for individuals, and no published context window sizes per plan. When any of these appear, this page and its date will change, and the change will be logged in our ChatGPT updates feed.