A solo realtor handling client details should choose Plus at $20 a month with training switched off, or Business from $40 a month for two seats if an assistant shares the work or the brokerage wants a no-training default. Free and Go can show ads, so keep them for work that carries no client information.
The plan changes how OpenAI treats your chats, not what your broker, your client agreements or fair-housing rules allow you to paste. Most realtors overspend on features they never open and underspend on the one setting that matters: whether client conversations can be used to improve the model.
The plans side by side for a solo brokerage
OpenAI sells individual plans (Free, Go, Plus, Pro) and a team plan (Business, which used to be called Team). Prices below are USD list prices as of November 2026; the pricing page shows local currency in some places, so check the figure at checkout before you commit.
| Plan | Price | Ads | Your chats and training | What a realtor gets |
|---|---|---|---|---|
| Free | $0 | May appear | May be used to improve models unless you switch it off in Data controls | Unlimited everyday text chats, limited uploads, limited image creation |
| Go | $8 a month | May appear | Same as Free; you can opt out | More messages with tools, more uploads, images and voice chats, longer memory, Projects |
| Plus | $20 a month | None | Same consumer rule: you opt out yourself | Expanded limits, deep research, Projects, scheduled tasks, ChatGPT Work |
| Business, Standard seat | $20 a seat annual, $25 monthly; two seats minimum | None | Not used for training by default | Shared workspace, Projects, apps, admin controls, central billing |
| Business, Premium seat | $100 a seat annual, $125 monthly | None | Not used for training by default | Five times the Standard seat's usage, with no five-hour limit; overkill for most realtors |
Pro sits far above what a realtor does in a day, so it is left out of the comparison. Go is the awkward middle: it costs money but keeps the ads and the training default of Free, which is why the sensible choices are usually Free for non-client experiments, Plus for one person, or Business for two or more.
Ads on Free and Go: why they matter in front of clients
OpenAI says ads are clearly labelled, sit apart from the answer and do not influence it, and that advertisers do not get your conversations. None of that makes an ad welcome on a screen you are sharing in a listing consultation.
Picture the moment: you cast ChatGPT to a seller's television to show how you will write their listing, and a sponsored card for another service sits under the draft. Nothing is wrong with the answer, but the seller has just watched a third party sell something in the middle of your pitch. Plus, Pro and Business accounts do not carry ads, which is worth the $20 on its own if you ever present from your account. You can switch off ad personalisation in Settings, but the ads themselves stay on the cheaper tiers.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowWhat the plan does not decide: client data by task
A paid plan does not turn every document into safe material. Sort tasks by what is inside them, not by how routine they feel. The table below is the rule I would give a solo brokerage; adjust it to your broker's policy and your MLS data licence, which can restrict how listing data is reused.
| Task | Free or Go | Plus | Business |
|---|---|---|---|
| Listing description from public listing facts | Fine | Fine | Fine |
| Open-house email and social posts | Fine | Fine | Fine |
| Buyer follow-up draft using first names only | Avoid | Fine with training off | Fine |
| Summary of an inspection report with names and addresses removed | Avoid | Fine with training off | Fine |
| Offer comparison with identifiers replaced by Offer A, B, C | Avoid | Fine with training off | Fine |
| Lead spreadsheet with phone numbers and emails | No | Avoid | Only if your broker allows it |
| Signed contract, ID or pre-approval letter | No | No | Ask your broker; redact first |
| Lockbox codes, alarm codes, banking details | Never | Never | Never |
The pattern is simple: public facts go anywhere, redacted facts go on a plan with training off, and identifiers or financial documents stay out unless a broker-approved tool handles them. If you want a longer treatment of the training question, how to stop AI tools training on your business data walks through the switches across vendors, and whether customer data is safe in ChatGPT covers the reasoning.
The cost sums for three common set-ups
Business needs at least two seats, which is where most solo realtors get the wrong answer. Here are the numbers for three illustrative brokerages, all on annual billing unless stated.
| Set-up | Plus route | Business route | Difference per year |
|---|---|---|---|
| Agent working alone | 1 x $20 = $20 a month, $240 a year | 2 x $20 = $40 a month, $480 a year (second seat idle) | Business costs $240 more |
| Agent plus part-time transaction coordinator | 2 x $20 = $40 a month, $480 a year | 2 x $20 = $40 a month, $480 a year ($600 if billed monthly) | Same on annual billing |
| Five-agent team | 5 x $20 = $100 a month, $1,200 a year | 5 x $20 = $100 a month, $1,200 a year ($1,500 monthly) | Same on annual billing |
Two things follow. For one person, Business is a $240 premium for a no-training default you can reproduce on Plus by flipping a switch. For two or more people, the prices match, so Business wins on the default, the shared workspace and a single invoice. The catch is commitment: seat cuts on Business have applied at renewal with no mid-term refund, so check the billing help page before buying five seats for a team that might shrink in March.
Quick payback check: at $20 a month, an agent who values an hour of admin at $50 recovers the subscription in 24 minutes. Almost any plan pays for itself on listing descriptions alone; the real question is which plan keeps you out of trouble.
Ten minutes to set Plus up safely
If you choose Plus, do these in order, once, and write the date in your policy notes.
- Open Settings, then Data controls, and switch off the option that lets conversations improve the model for everyone. OpenAI labels it "Improve the model for everyone", so check the wording on your screen.
- Review memory. Decide whether you want ChatGPT remembering your tone and territory, and remove anything that names a client. Memory settings change from time to time, so describe your own choice in your policy note rather than relying on a screenshot.
- Use Temporary Chat for anything sensitive. OpenAI keeps those chats for up to 30 days, so they are not a place for documents you could never share.
- Create a Project called "My voice and my area", paste the instructions below as project instructions, and upload three of your best listing descriptions as examples. Account-wide custom instructions allow up to 5,000 characters on Plus but only 1,500 on Free and Go.
- Turn on two-step sign-in. A stolen realtor login is a stolen client history.
- If a payment lapses and the account drops to Free, check the ad settings and re-check the training switch.
Custom instructions that work for most realtors, kept short so they leave room for the task:
You write for a solo real estate agent. Voice: warm, plain, specific. No exclamation marks, no "stunning" or "must-see".
Describe the property and its features, never the type of buyer who should live there.
Never comment on school quality, crime, religion, family status, age, disability or ethnicity of an area or its residents.
Use only the facts I supply. If a fact is missing, ask instead of inventing it.
Never ask for or repeat names, phone numbers, addresses of clients, ID numbers or financial details.Three prompts with sample output and what to fix
These are illustrative, so you can see how the setting above shapes the work. The first draft is rarely what you send.
A listing description from public facts
Write a 120-word listing description. Facts: 3 bed, 2 bath, 1,640 sq ft, renovated kitchen 2024, south-facing garden, single garage, 8 minutes' walk to the park, asking $449,000. Tone: warm, no hype.A typical first draft opens with "Perfect for young families, this stunning home sits near the best schools." Two problems: "perfect for young families" and "best schools" both steer buyers by family status and neighbourhood, which fair-housing rules treat with suspicion. Rewrite it to describe the property: "A bright three-bedroom home with a renovated 2024 kitchen, south-facing garden and single garage, eight minutes' walk from the park." The custom instructions above stop most of this, but read every draft for it anyway.
A buyer follow-up with placeholders
Draft a 90-word follow-up to a buyer after a Saturday showing. Use [first name]. Mention the kitchen and the commute they asked about. Offer two time slots for a second visit. Do not mention price negotiation.The draft will usually be fluent and a little too eager ("I'm thrilled to hear what you thought!"). Cut the enthusiasm, add the two slots, and check that it commits you to nothing about condition or price. You paste the real name in your email client, not in ChatGPT.
An inspection summary after redaction
Before: the report header lists the seller, buyer, property address and the inspector's licence number. After: replace them with [seller], [buyer], [address] and [inspector], keep the findings, then ask for a one-page summary sorted into safety, structural and cosmetic items. Sample output: "Safety: loose handrail on the rear steps; two ungrounded outlets in the garage. Structural: hairline crack in the north foundation wall, monitor. Cosmetic: worn caulk, cracked bathroom tile." Your fix: verify each line against the report, because summaries drop qualifiers such as "recommend evaluation by a specialist", and that qualifier is often the whole point.
The pre-approval letter pasted into a Free account
This is an illustrative scenario. An agent on the Free plan pastes a buyer's pre-approval letter and bank name into a chat to ask how to word a counteroffer. The reply is fine. Three weeks later, the agent's broker asks how client financials are handled and discovers the chat history holds the letter on an account with training on. Nothing leaked in public, but the agent cannot say where the letter went, cannot be sure it is deleted everywhere, and has to tell the buyer. The cheap fix would have been a rule (financial documents never go in) plus Plus with training off. The better fix was a Business workspace with a written boundary and an admin who can see who is using what.
What Business adds beyond the no-training default
For a two-person team the extras are modest but real. OpenAI says it will not train on a Business workspace's data, the workspace has admin controls and central billing, and it includes Projects, apps (formerly called connectors) and Company Knowledge. ChatGPT Work, the agent mode that replaced ChatGPT agent in July 2026, is on Standard seats too.
Know the limits as well. A Business workspace has no self-service data export, so you cannot easily leave with your history. Invoicing and custom data-retention policies sit on Enterprise, not on the self-serve Business plan. Business is capped at 200 paid seats, which will not trouble a brokerage but is worth knowing if you manage a large team. And no plan includes API use, which is billed separately.
If your assistant works inside your CRM rather than in ChatGPT, read what goes to a virtual assistant and what goes to AI before you add a seat for them. A second seat is wasted if the assistant never opens ChatGPT.
A working week mapped to the plans
A plan only earns its price if it fits the jobs you repeat. Here is an illustrative week for a solo agent with six live listings, with the minutes before and after drafting with ChatGPT, and the data rule each job triggers.
| Day and job | Minutes by hand | Minutes with a draft | Data rule it triggers |
|---|---|---|---|
| Monday: four listing descriptions | 100 (25 each) | 32 (8 each) | Public facts only, any plan |
| Tuesday: twelve lead follow-ups | 72 (6 each) | 25 | First names added in your email client; plan with training off |
| Wednesday: one inspection summary | 40 | 12 | Redact first; Plus with training off or Business |
| Thursday: three-offer comparison | 45 | 15 | Offer A, B, C instead of names; same rule |
| Friday: neighbourhood market newsletter | 90 | 35 | Public figures; check steering language |
| Total | 347 minutes | 119 minutes | Saves about 3.8 hours |
At $50 an hour, 3.8 hours is $190 of time in a week against $20 a month for Plus, so the subscription is not the real decision. The decision is that three of the five jobs touch material that should not sit on a plan with training on, which is exactly where Free and Go fail you. If your week is mostly Monday and Friday, Free handles it and you can wait. If Tuesday to Thursday is your week, Plus is the floor.
The same table shows why more seats rarely pay off early. A transaction coordinator who only drafts reminders and chases signatures gains less from an assistant than the agent does, so give them a seat once their weekly list contains at least two jobs like the ones above.
Questions to put to your broker before you upgrade
Many brokerages have a technology policy that predates AI. Ask four things in writing and keep the answers in one file.
- Does the brokerage allow generative AI on client work, and is there a list of approved tools or plans?
- Is client information, including names and addresses in transaction files, allowed to leave the brokerage's own systems for a third-party service?
- Must AI-written marketing be reviewed or disclosed, and who signs it off?
- Does any MLS licence or listing-data agreement limit using listing data in tools like this?
If the answer to the second question is no, the plan comparison becomes moot for client documents: you use ChatGPT only for public-facts work and handle the rest in your transaction system. That is a perfectly good outcome, and it costs nothing.
A paragraph you can adapt for your file, so the rule exists before anyone needs it:
I use ChatGPT Plus (training switched off) or a Business workspace to draft marketing copy and to summarise redacted documents. I do not enter client names, contact details, financial documents, signed contracts or access codes. I read every draft before it is sent or published, I do not use AI to describe the type of person who should buy a property, and I review this note each quarter.
A four-question rule for choosing
- Do you ever present from your own account in front of clients? If yes, skip Free and Go, because of the ads.
- Does anyone else need to use it? If yes, choose Business: it costs the same as separate Plus accounts on annual billing, and each person gets their own login as the terms require.
- Will you paste redacted client material? If yes, choose Plus with training off, or Business, never Free.
- Does your broker require a written no-training commitment? If yes, choose Business and keep a copy of the Business Terms in your files.
If your answers are no, no, no, no, Free is enough for listing copy and open-house posts. The moment one answer flips, pay. For help picking between assistants, Claude or ChatGPT for realtors compares them on the daily jobs, and whether a small business needs an AI policy gives you the one-page rule to write down.
What to re-check every quarter
Plans change quickly. Once a quarter, spend ten minutes on four things: the price on the pricing page, whether ads have reached your tier, whether the training default has changed, and which features moved between plans. Model names, usage limits and feature names rotate often, so write "ChatGPT" in your policy rather than a version. Custom GPTs are the example to remember: they were part of the plans and are now being retired, so anything you built there needs a Project home before 11 December 2026.
If you would rather map this with someone, that is what my AI implementation consultation covers, starting from the jobs you actually repeat each week; get in touch when you are ready.
Plan questions realtors ask
Does ChatGPT Plus show ads?
No. OpenAI says ads may appear on the Free and Go plans (its pricing page labels only Go), and that Plus, Pro, Business and Enterprise accounts carry no ads. If you ever screen-share ChatGPT with a client, that is a practical reason to avoid the cheaper tiers.
Can my assistant and I share one Plus login?
No. OpenAI's terms ban sharing account credentials, so each person needs their own login. Two Plus subscriptions cost $40 a month, the same as two Business seats on annual billing, so Business is the better fit once a second person is involved.
Do I need Business to use Projects?
No. Projects work on every plan, Free included, and OpenAI says project sharing does too; Go and Plus projects hold 25 files against 40 on Business. What Business adds is the no-training default, central billing and admin controls, not the Projects feature itself.
Should I build a custom GPT for my listing voice?
No. OpenAI has ended custom GPT creation and existing GPTs stop running on 11 December 2026. Put your voice rules and example listings in a Project instead, which carries the same instructions and files.
Is ChatGPT Business enough to paste a signed contract?
The plan decides how OpenAI treats your data, not whether your brokerage, your client agreement or your local rules allow the upload. Ask your broker first, redact identifiers anyway, and keep financial documents out entirely.
Further reads
- Best AI CRM for Real Estate Agents: Follow Up Boss, Lofty, BoldTrail — Where lead follow-up should live once the assistant is chosen.
- AI Receptionist for Realtors: Cost, Call Routing and Showing Cover — Cover showing calls without paying for another assistant seat.
- AI Lead Generation for Real Estate Agents: No Bought Lists Needed — Use the assistant for lead work that avoids bought lists.
- AI Real Estate SEO: Neighbourhood Pages That Avoid Steering — Write area pages without drifting into steering language.
- Does ChatGPT Remember Conversations? Memory Settings for Business — Memory settings explained for business accounts.
- How to Classify Business Data Before Using AI Tools — A simple way to sort what may and may not be pasted.
- AI Virtual Staging for Property Listings: Costs and Disclosure — Costs and disclosure when listing photos are staged with AI.
- Follow Up Boss AI: Which Plan Adds Smart Summaries and Suggested Tasks? — All three Follow Up Boss plans list smart summaries and suggested tasks, but the AI leans on calling data. Plan-by-plan features and when Pro beats Grow.
- Rechat AI for Brokerages: What Lucy Does and What It Costs — Lucy's features from Rechat's own help pages, the pricing Rechat does not publish, and a 30-day pilot plan for a ten-agent office.
- AI Tools for Real Estate Brokers: From Transactions to Agent Support — Five tool types for a small brokerage, each with published costs and watch-outs, a worked cost for 12 agents and a 90-day adoption plan.
- Best Free AI for Realtors: Three Jobs, the Limits and When to Pay — Three jobs a realtor can do on free AI plans, a fair-housing check for every output, and the signs that a paid plan is worth it.
- ChatGPT Bio for a Real Estate Agent: Team Standards, Prompts and Edits — Team standards, an interview prompt, three before-and-after bios and a claims check for agent bios written with ChatGPT.
- ChatGPT Training for Real Estate Agents: A One-Week Team Plan — A team leader's one-week plan for training agents on ChatGPT: daily sessions, a rules sheet, five workflows and an adoption check.
- ChatGPT for Realtors: Brokerage Rules for Emails, Disclosures, Follow-Ups — A brokerage owner's rulebook for agent ChatGPT use: what may be summarised, what must be redacted, the fact-check pass and five reusable prompts.
- Real Estate Agent Recruiting With AI: Outreach and Onboarding Plan — A four-touch recruiting sequence, an onboarding checklist and a first-30-days plan for a brokerage, with the compliance lines AI must not cross.
- Can Realtors Use HeyGen Avatars for Listings? Rules and Setup — How a realtor can put their own HeyGen avatar on listing updates, what to disclose, and how to avoid MLS, brokerage and fraud trouble.
- ChatGPT Ads for Realtors: Access, Cost and Fair-Housing Rules — What a realtor may and may not advertise on ChatGPT, what a first month costs, and the fair-housing checks for ad copy and targeting.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: OpenAI pricing page, ChatGPT Business overview, Ads in ChatGPT and ChatGPT Go help articles, OpenAI business data page (checked 1 November 2026).