Three questions, and a reason next to every answer. Every price and limit here shows its source and the date we checked it.
Mistral AI
$14.99/month
Why: A full assistant at $14.99, the lowest official price among the general assistants.
The lowest-priced full assistant here, and the $15 of API credits nearly equals the subscription if you also build with the API. "More messages" has no published number.
See limits and details
Anthropic
$20/month
Why: $20 a month, or $17 billed annually. No message count: a five-hour window plus a weekly limit.
There is no fixed message count. A long session can use up the five-hour window, then you wait for it to reset; the weekly limit caps heavy use across the week. Claude Code draws from the same allowance.
See limits and details
OpenAI
$20/month
Why: $20 with Deep Research and Agent mode. Price and limits from a third-party source.
Enough for a working day of normal chat for most people. The number comes from a third-party article, and OpenAI has changed it before without notice.
See limits and details
Every price and limit here shows its source and the date we checked it. No affiliate links: the order comes from the table, not from who pays.
All 24 rows the quiz can return, in the order it reads them. Nothing else decides the result.
| Use | How often | Plan | Price | Why |
|---|---|---|---|---|
| Writing | a few times a week | ChatGPT Go | $8 | The lowest paid price in the table ($8). Price from a third-party source. |
| Writing | a few times a week; every day | Le Chat Pro | $14.99 | A full assistant at $14.99, the lowest official price among the general assistants. |
| Writing | a few times a week; every day; all day, for work | Claude Pro | $20 | $20 a month, or $17 billed annually. No message count: a five-hour window plus a weekly limit. |
| Writing | every day | ChatGPT Plus | $20 | $20 with Deep Research and Agent mode. Price and limits from a third-party source. |
| Writing | all day, for work | Claude Max | from $100 | From $100 for 5x the usage of Pro, with a 20x option. |
| Writing | all day, for work | ChatGPT Pro | from $100 | From $100 for 5x the usage of Plus. Third-party source. |
| Coding | a few times a week; every day | Copilot Pro | $10 | $10 with unlimited completions and 1,500 AI credits a month: the cheapest paid coding plan here. |
| Coding | a few times a week; every day | Le Chat Pro | $14.99 | $14.99 with a coding agent for CLI and IDE, plus $15 a month in API credits. |
| Coding | every day; all day, for work | Claude Pro | $20 | $20 and it includes Claude Code, which draws from the same allowance as chat. |
| Coding | every day | Cursor Pro | $20 | $20 for an AI editor with agents. Usage beyond the included amount is billed on demand. |
| Coding | all day, for work | Copilot Pro+ | $39 | $39 for 7,000 AI credits a month. |
| Coding | all day, for work | Cursor Pro+ | $60 | $60 with more included usage than Cursor Pro. |
| Coding | all day, for work | Claude Max | from $100 | From $100 for 5x the Pro allowance, which is what long Claude Code sessions consume. |
| Coding | all day, for work | Copilot Max | $100 | $100 for 20,000 AI credits a month. |
| Coding | all day, for work | Cursor Ultra | $200 | $200 with the largest included usage of the individual Cursor plans. |
| Research | a few times a week; every day; all day, for work | Perplexity Pro | $20 | $20 for a search-first assistant whose answers cite sources. Price from a third-party source. |
| Research | a few times a week; every day | Google AI Pro | $19.99 | Deep Search plus 5 TB of storage. Official price in Brazil: R$ 96.99. |
| Research | every day; all day, for work | Claude Pro | $20 | $20 and the Research feature is included. |
| Research | every day | ChatGPT Plus | $20 | $20 with Deep Research tasks. Third-party source. |
| Research | all day, for work | ChatGPT Pro | from $100 | About 250 Deep Research tasks a month on the $200 tier. Third-party source. |
| Research | all day, for work | Perplexity Max | $200 | $200 for agent features on top of Pro. Third-party source. |
| Images | a few times a week; every day | Google AI Pro | $19.99 | Video generation and 1,000 Flow credits a month. Official price in Brazil: R$ 96.99. |
| Images | a few times a week; every day; all day, for work | SuperGrok | $30 | $30 with one weekly pool shared between chat and Imagine. Third-party source. |
| Images | all day, for work | Google AI Ultra | R$ 779.90 | 10,000 to 25,000 Flow credits a month. We only have its price in Brazil (from R$ 779.90). |
The quiz keeps the rows for your use and frequency, drops the ones above your budget (Up to $10, Up to $20, Up to $100 or none) and shows the first three in this order.
A paid plan buys three things: a higher limit, access to the stronger models, and extra tools such as research or coding agents. If you never reach the limit of the free version, the first one is worth nothing to you, and the question becomes whether you need the other two.
A simple test: use the free version for two normal weeks and count how many times it stopped you. Zero or one means stay free. A few times a week means a $20 plan. Stopped on the $20 plan most days means the $100 step, and only then.
It does not ask which tools your workplace already pays for, which language you write in, or whether you care about a specific feature. Those can outweigh everything in the table. Treat the result as a short list to check against the plan pages, not as a verdict.
A score hides its reasoning. A row in a table can be read, disputed and corrected. When a provider changes a price or a limit, the row changes and the change is visible in the project history.
It filters a written table. Each row says: for this use and this frequency, this plan, for this reason. The quiz keeps the rows that match your answers, removes the ones above your budget and shows the first three in the order of the table. The whole table is printed below the quiz.
Because it is only the first row of the table that matches. Whether it is best for you depends on things the quiz does not ask, such as which tools your team already uses. The reason shown next to each plan is what you should weigh.
No. The order is fixed in the table and the site has no affiliate links or paid placements.
The quiz says so and shows the cheapest row for your use instead of pretending a plan fits.
The data covers eight services so far. Dedicated image and video tools are not in it yet, so the quiz cannot return them. Adding a service means someone has to re-check it regularly, so the list grows slowly on purpose.