Claude and ChatGPT just got more generous. Here is the August 2026 pricing table, and what a real estate firm should actually buy.

Both major assistants moved this year in the same direction: more usage for the same money, and a cheaper way in at the bottom. If you priced these tools six months ago and decided they were too expensive or too limited, the math has changed.

Here is every current plan, what actually changed, and the part nobody writes down: which one a small real estate firm should put on the company card.

What changed on usage

The bigger story this year is not price, it is how much you get for it.

On May 6, Anthropic doubled Claude Code’s five-hour rate limits across Pro, Max, Team and seat-based Enterprise, and removed the peak-hours reduction that used to throttle Pro and Max accounts in the busy part of the day. The capacity behind it came from an agreement to use SpaceX’s Colossus 1 data center, over 300 megawatts and more than 220,000 GPUs.

The practical version: the ceiling most people used to hit on a heavy afternoon moved. If you tried Claude in early 2026, ran into limits, and wrote it off, that specific complaint is largely fixed.

OpenAI moved at the other end, adding a Go tier at $8 a month. That is the first time either company has put a paid tier meaningfully below $20, and it matters for firms who want to give a whole office access without buying everyone a full seat.

The table, as of August 2026

Claude

Plan	Monthly	Annual	What you get
Free	$0	$0	Chat, web search, file creation. No Claude Code.
Pro	$20	$17/mo	All models, plus Claude Code, Cowork, Design and Science in one shared pool
Max 5x	$100	$100	5x the Pro usage
Max 20x	$200	$200	20x Pro usage
Team Standard	$25/seat	$20/seat	Team pool, admin controls
Team Premium	$125/seat	$100/seat	5x the standard pool
Enterprise	Custom	~$20/seat + API	Compliance controls, data residency

ChatGPT

Plan	Monthly	Annual	What you get
Free	$0	$0	27K context, limited messages
Go	$8	n/a	54K instant / 256K reasoning context
Plus	$20	n/a	Same context as Go, plus the named GPT-5.6 models, deep research, Projects
Pro	From $100	n/a	128K instant / 400K reasoning, 5x to 20x more usage
Business	$25/user	$20/user	Minimum 2 users, SSO, admin console, spend controls
Enterprise	Custom	Custom	SCIM, data residency across 10 regions

Two things worth noticing in that table. At $20 the plans look identical and are not: Claude’s Pro bundles its agent products into the same subscription, while ChatGPT’s equivalent is mostly about model access and context. And OpenAI’s $8 Go tier has no real counterpart on the Claude side, so if your constraint is seats rather than depth, that gap is the story.

About those promo codes

We went looking for current promotions so you would not have to, and the honest answer is that almost everything you will find is noise.

Search either product with the word “coupon” and you get pages promising 50%, 75%, even 35% off, from sites that make money on the click. Neither company runs public discount codes on consumer subscriptions. Treating those pages as real is how people end up on resold or shared accounts, which is a genuinely bad idea for a firm putting client information into a chat window.

The real discounts are unglamorous and they are on the pricing page:

• Annual billing. Claude Pro drops from $20 to $17 a month, Claude Team from $25 to $20 a seat, ChatGPT Business from $25 to $20 a user. That is 15 to 20 percent for paying up front.
• Start lower than you think you need. Both companies let you move up mid-cycle. Buying Max or Pro on day one, before you know your actual usage pattern, is the most common way firms overspend here.

If you see a genuine promotion from either company directly, it will be on their own site. Anywhere else, assume affiliate.

What we would actually buy

For a small brokerage or property management firm, the pattern that has worked for us and for the operators we advise:

One seat at $20, not five. Pick the person whose week has the most repeatable document work, usually operations or the person assembling owner reports, and give them a full seat first. Give everyone else free access to prove out whether the demand is real before you multiply the spend.

Go up only when you hit a wall you can name. “We ran out of usage twice last week while doing the month-end pack” is a reason to move to Max. “It feels slow” is not.

Annual only after 60 days. The 15 percent saving is real, and it is not worth locking in before you know whether the tool fits your workflow. We have watched firms annualize in week one and abandon the tool in week six.

The uncomfortable truth in all of this is that the subscription is the cheap part. At $20 a month, the tool costs less than a single hour of the work it is meant to replace, which means the decision was never really about price. It is about whether you have identified a process worth automating and whether anyone owns making it stick.

If you want help finding that process before you spend anything, that is a 30 minute conversation.

Sources: Anthropic on higher limits and the SpaceX compute deal, CloudZero’s Claude pricing breakdown, Digital Applied’s ChatGPT plan comparison, August 2026. Prices as published at the time of writing and changing frequently. Check the vendor’s own pricing page before you buy.