If you pay for an AI assistant, you have probably looked at the $100 and $200 plans and wondered whether they are worth it. Here is a pattern that most pricing pages do not say out loud. On the multiples reported as of September 30, ChatGPT's paid tiers all worked out to the same $20 per unit of usage relative to the $20 Plus plan. Anthropic's Claude plans follow the same ladder today. Once you see that, the question stops being "which plan is cheapest" and becomes "how much usage do I actually burn." This guide walks through Anthropic's current pricing page, the September 30 ChatGPT snapshot, what OpenAI changed on October 29, and the traps that make the simple math less simple than it looks.
Quick summary: ChatGPT Plus is $20 (1x), Pro 100 is $100 (5x), Pro 200 is $200 (10x) and Pro 500 is $500 (25x), per a tracker citing OpenAI's help pages as of September 30. That was $20 per unit on every tier. Claude Pro is $20 (1x), Max is $100 (5x) and Ultra is $200 (10x weekly), per Anthropic's pricing page, so higher Claude tiers buy headroom, not a discount per unit. On October 29, OpenAI said it was doubling usage limits on its Pro tiers, so the ChatGPT rows below are a September 30 snapshot, not current multiples. OpenAI has not published the new numbers. Prices exclude tax.
The ladder, side by side
| Plan | US price per month | Usage vs the $20 plan (ChatGPT: Sept. 30 snapshot, before the Oct. 29 doubling) | Price per usage unit |
|---|---|---|---|
| ChatGPT Go | $8 | More than Free, below Plus (no multiple published) | Not comparable; may include ads |
| ChatGPT Plus | $20 | 1x | $20 |
| ChatGPT Pro 100 | $100 | 5x | $20 |
| ChatGPT Pro 200 | $200 | 10x (was 20x before a Sept. 29 cut) | $20 |
| ChatGPT Pro 500 | $500 | 25x | $20 |
| Claude Pro | $20 monthly, or $200 up front for a year | 1x | $20 monthly, about $16.67 on the annual plan |
| Claude Max | $100 | 5x Pro usage | $20 |
| Claude Ultra | $200 | 10x Pro usage weekly, no 5-hour limit for verified accounts | $20 |
The ChatGPT multiples come from Nerd's Chalk, which says its table was updated on September 30, 2026 and relies on OpenAI's help pages. A search of OpenAI's help center, which we could not open directly, returned the same 5x, 10x and 25x relationships. The Claude numbers come straight from Anthropic's pricing page: Pro at $20 billed monthly or $200 billed up front, Max at $100 with 5x Pro usage every 5 hours and every week, and Ultra at $200 with no 5-hour limit for verified accounts and 10x Pro usage every week. The price per unit column is our arithmetic. (Sources: Nerd's Chalk, Mixed News, Anthropic)
Why the Pro 200 line is the interesting one
Before September 29, Pro 200 was reported as 20x Plus, which meant $10 per unit, a better deal than Pro 100. On September 29, OpenAI added the $500 tier and reopened Pro 200 sign-ups with a reduced allowance at the same $200 price, citing "our increasingly efficient models," per Mixed News. Nerd's Chalk reports that sign-ups had been paused since September 10 and that existing subscribers kept the old allowance through October 29. Then came backlash. In an October 29 post relayed by BleepingComputer, an OpenAI representative wrote, "We heard your kind feedback and won't be reducing usage limits for Pro 200," and said all usage limits for the Pro 100, 200 and 500 tiers and Business Premium are doubling effective immediately. (Sources: Mixed News, Nerd's Chalk, BleepingComputer)
That sequence is the first catch. Usage multiples are a policy, not a law of nature. A provider can change them with a release note, and in this case two changes landed within a month. BleepingComputer's report does not give exact limits after the doubling. Later coverage from Online Tech Tips says the increase applies to Work and Codex usage on Pro 100 and Pro 500, that Pro 200 subscribers on the reduced allowance were moved back to the higher one, and that nothing indicates Plus changed. So the clean $20-per-unit table above is a September 30 snapshot, not current pricing. If the doubling applies on top of those multiples, the effective cost per unit on Pro tiers would be roughly half what the table shows, but OpenAI has not published the new multiples. Check OpenAI's help page before you buy. (Source: Online Tech Tips)
Five catches in the simple math
- Units are relative to each company's own base plan. A 5x on ChatGPT is five times Plus, and a 5x on Claude is five times Pro. Those two baselines are not the same amount of work, so you cannot compare $100 plans across vendors by multiple alone.
- Usage depends on what you run. Heavy reasoning, long files, coding agents and image or video generation burn allowance faster than short chats. Anthropic's page says limits vary by model, chat length and file size.
- Limit windows differ. Claude's tiers list limits every 5 hours and every week. Nerd's Chalk says Plus has a five-hour limit while Pro 100 has none. A weekly cap and a five-hour cap feel very different in practice.
- Some tiers are mostly about features, not usage. Nerd's Chalk reports that only Pro 500 includes Ultrafast, a faster service tier. Anthropic lists $100 and $200 in monthly API credits on Max and Ultra, respectively, which changes the value if you also build with the API.
- Cheaper tiers may carry ads. MacRumors reported in January that OpenAI planned to test ads in the free and Go tiers for logged-in adult U.S. users, with Plus, Pro, Business and Enterprise staying ad-free.
A practical way to choose a tier
Because Claude's tiers cost the same per unit, and ChatGPT's did until the October change, the cheapest tier that never stops you is usually the right one. Overpaying for headroom you do not use is the real way to waste money here. A simple method works for most people.
- Start at $20. Use it normally for two to four weeks and note every time you hit a limit message and how long you waited.
- If you hit a limit once or twice a month, stay. The wait is cheaper than a $80 monthly jump.
- If you hit a limit several times a week during paid work, price the lost time. If waiting costs you more than $80 a month in hours, the $100 tier pays for itself. If it does not, stay at $20.
- Go to $200 only if you hit the $100 ceiling regularly. At that point you are almost certainly running long agent or coding sessions.
- Treat $500 as a niche tier. Per the sources above, the main differences beyond raw allowance are Ultrafast and the Agents API with computer use.
- Consider the annual option where it exists. Anthropic lists Pro at $200 billed up front, which is about 17 percent less than twelve months at $20, per our arithmetic.
Not financial or tax advice. If you pay for AI through a business, whether the cost is deductible or reimbursable depends on your situation. Ask a qualified tax professional or your employer's finance team.
What the $100 tier was designed to do
Engadget reported on April 10, 2026 that OpenAI added a $100 Pro plan between $20 Plus and $200 Pro, positioned against Anthropic's $100 option. That article says the $100 plan offered five times the Codex of the $20 plan, with a limited-time promotion that doubled it, and that the $100 and $200 plans include the same advanced tools and models, with the difference being usage limits. An OpenAI spokesperson told TechCrunch, as quoted by Engadget, "Compared with Claude Code, Codex delivers more coding capacity per dollar across paid tiers." That is a company claim, and we have not independently tested it. The structural point stands: the $100 tier exists because many heavy users wanted something between $20 and $200. (Source: Engadget)
The bigger pricing trend
Looking across the table, the ladder has converged: a $20 entry plan, a $100 power-user plan, a $200 heavy plan, and in OpenAI's case a $500 top plan. That is not a coincidence, since both companies price against each other. It also means the real competition is on allowance size and on how often the allowance rules change. Watch release notes, not just price tags. A plan that stays at $200 while its allowance moves is, in effect, a price change.
If you do not want to commit to a $100 monthly plan to find out, LumiChats offers a pay-per-day option and lets you try several AI assistants in one place. That is a low-risk way to learn how much AI you actually use before you lock in a tier. Our earlier posts on whether ChatGPT Plus is worth it and how LumiChats pricing works have more on that comparison.
01Do higher ChatGPT tiers give a better price per unit of usage?
Not on the multiples published as of September 30: $20 for Plus (1x), $100 for Pro 100 (5x), $200 for Pro 200 (10x) and $500 for Pro 500 (25x) all equal $20 per unit. OpenAI announced on October 29 that usage limits on its Pro tiers were doubling, so current multiples differ and per-unit costs on Pro are likely lower. OpenAI has not published the new figures.
02How much is Claude Max and what does it include?
Anthropic's pricing page lists Max at $100 with 5x Pro usage every 5 hours and every week, plus $100 per month in API credits. Ultra is $200 with 10x Pro usage every week and $200 in API credits.
03Did OpenAI cut the Pro 200 allowance?
It reduced the allowance for new subscribers on September 29 at the same $200 price, according to Mixed News. On October 29 it said it would not reduce limits for Pro 200 and would double limits across Pro tiers, per BleepingComputer.
04Is ChatGPT Go worth it at $8?
Go is $8 per month in the U.S., per MacRumors, with more messages, uploads and image creation than Free. It may include ads, and the more advanced reasoning models require Plus or Pro as of that January report. It suits light users, not heavy ones.
05Can I compare a $100 ChatGPT plan with a $100 Claude plan by multiples?
No. Each multiple is relative to that company's own $20 plan, and usage depends on model, task and file size. Test both on your real work if you can.
06Are these prices final?
No. Prices exclude tax and are subject to change, and ChatGPT allowances changed twice between September 29 and October 29. Check each provider's current pricing page before paying.
