ChatGPT for Real Estate Agents: Best AI Tools by Job

Which AI assistant fits each real estate job?

ChatGPT and Claude are our starting recommendations for drafting; no assistant replaces professional review. These are editorial preferences by job, not an overall model ranking or a claim that a particular subscription is required.

Prices and plans checked 2026-10-19 on the vendors’ official pages. All prices below are USD per month, excluding tax.

The plan column is a purchasing starting point, not a verified minimum requirement. Our API results cannot tell you whether a free consumer plan will meet your workload. Gemini 4 Argon passed 88% of criteria on our round-1 real estate prompts, graded by GPT-6 Astra, but market summaries and Copilot remain untested. Consumer-app results may differ from these controlled tests.

For broader workflows, see our guide for real estate agents. Readers in the UK, Australia and India should check local prices, taxes and professional rules, which differ from the US.

If you need a paid plan, compare billing terms before subscribing:

Prices and plans checked 2026-10-19 on official vendor pages; USD list prices excluding tax.

Claude Free includes Sonnet and Haiku, not Opus or Fable; Pro includes Opus, Sonnet and Haiku, not Fable. OpenAI’s October system card announces GPT-6 across free and paid ChatGPT plans, but the default model split by plan is not verified. These distinctions matter when comparing subscriptions with API test results.

What did our real estate tests show?

Listing and follow-up prompts met every rubric criterion in the tested runs, while offer summaries did not. That is evidence about these prompts under controlled conditions, not proof of dependable performance across transactions.

Scores below come from a command-line API harness with tools, web and memory off, except Gemini 4 Argon, which ran through Google’s Antigravity CLI with a tool-less agent; none came from consumer apps, so app results may differ. Judges were not equally strict: Claude and Gemini outputs were graded by GPT-6 Astra, and GPT outputs by Claude Opus 5.5. Part of the cross-family gap comes from the grader, so these scores do not establish an overall winner.

Test date: 2026-10-19. Each task cell shows average criteria passed out of 5, followed by runs meeting every criterion out of 3.

Claude Fable 5.1 and Gemini 4 Argon each met every criterion in 3 of 3 runs on both the follow-up and listing prompts; on the offer summary, they scored 3.33 and 3 of 5 respectively, each with 0 of 3 runs meeting every criterion. Across round-1 real estate prompts they passed 89% and 88% of criteria respectively, against 95% for Opus and 92% for Sonnet under the same judge, GPT-6 Astra.

In round-1 calibration, swapping judges on 30 sampled Claude outputs changed the criteria pass rate from 93% to 98%. For 30 sampled GPT outputs, it stayed at 97%. Opus, Sonnet, Fable and Gemini shared the GPT-6 Astra judge, making comparisons among them like-for-like. An editor agent, not a human, independently checked 20 verdicts for the initial three models and agreed with 18.

“Tested” means at least 2 runs completed and graded, not professional sign-off. On 2026-10-19, round 1 covered 40 prompts across 10 professions and 360 runs. Claude Fable 5.1 and Gemini 4 Argon were added on 2026-10-20 using the same prompts and rubrics across all 80 profession prompts, with 240 runs per added model completed and graded. Read the test-round report and the methodology for the design and grading rules. The practical takeaway is to budget review time even when a draft reads smoothly.

How do I keep listing copy within Fair Housing guardrails?

Describe the property, not the people who should live there, and review both the copy and the way the advertisement is distributed. ChatGPT or Claude can draft from verified notes; neither can certify compliance.

HUD’s Office of Fair Housing and Equal Opportunity guidance dated April 29, 2024 explains how the Fair Housing Act applies to housing advertising through digital platforms, including automated targeting and delivery. NAR Code of Ethics Article 12 requires REALTORS® to present a true picture in advertising and other public representations.

This prompt is adapted from our tested listing prompt. The source-of-income instruction is a drafting precaution, not a statement that all listed protections have identical legal treatment everywhere.

Write listing copy for a US residential property from the notes below.
Deliver:
- An MLS public remarks paragraph of at most 250 words.
- A social caption of at most 40 words.
- A "Removed or changed" list explaining each omitted or revised phrase.
Rules:
- Describe the property, not the people who should live there.
- Avoid preferences or limitations based on race, color, religion, sex,
  disability, familial status or national origin.
- Also avoid source-of-income limitations.
- Use only verified facts in the notes.
- Flag uncertain claims and anything the listing broker should check.
Notes:
[VERIFIED PROPERTY FACTS AND SEPARATELY LABELED UNVERIFIED CLAIMS]

Compare every measurement, upgrade and amenity with your supporting records. Read the removal list: it can reveal questionable wording, but it is not protection against an omission or a mistaken explanation.

Ask your broker to review applicable local rules and the platform’s targeting and delivery settings. Reviewing the text alone does not address the full scope of HUD’s digital-advertising guidance. For reusable instructions, see our compliant-listing Claude Skill.

How should I draft follow-ups, captions and market summaries?

Give the assistant verified source material, a defined output and explicit instructions about missing information. ChatGPT or Claude are reasonable starting points for follow-ups and captions; use an existing assistant for market summaries before buying another subscription.

For follow-ups, separate each lead’s notes and preserve their contact preferences. This prompt is adapted from our tested follow-up prompt:

Write open-house follow-ups for each lead below:
one SMS of at most 320 characters and one email of at most 120 words.
Rules:
- Reference something specific the lead said. No pressure tactics.
- Respect stated contact preferences; mark any message that should not
  be sent under those preferences.
- Do not add prices, mortgage rates or property facts absent from the notes.
- Do not characterize neighborhoods or schools as good or bad,
  or describe who lives there. Point to objective sources instead.
Leads:
[ANONYMIZED LEAD LABELS, COMMENTS, CONTACT PREFERENCES AND CONSTRAINTS]

Check that each message belongs to the right lead and that no missing detail has become an assertion. A polished tone is not evidence that the details are correct.

For captions, use ChatGPT or Claude with the same property-focused rules. This standalone adaptation was not separately tested:

Write a social caption of at most 40 words using only the facts below.
Describe the property, not its intended occupants.
Do not add neighborhood, school, price or amenity claims.
Avoid preferences or limitations based on protected characteristics.
Separately flag any phrase or fact the listing broker should review.
Facts:
[VERIFIED PROPERTY FACTS]

Review the caption independently of the longer listing. Shortening copy can remove a qualification that matters. More reusable wording is available in our prompt library.

Market summaries were not tested. Gemini or Copilot are options if you already use them; that recommendation does not assume access to your files or an integration. Paste an approved source table and request a bounded summary:

Summarize the supplied local market data for a seller newsletter.
Rules:
- Use only figures in the data. Do not estimate or add outside figures.
- State the source and reporting period in the opening sentence.
- Do not predict prices or mortgage rates.
- Do not characterize neighborhoods or their residents.
- Flag unclear labels, missing values and uncertain interpretations.
Data:
[SOURCE, REPORTING PERIOD AND APPROVED MARKET DATA TABLE]

Check every figure against the table, including the reporting period and what each column measures. Do not let the assistant turn a limited dataset into a claim about the whole market.

Before supplying any client, lead or transaction information, read the plan’s own data/privacy terms and your professional confidentiality rules. We have not verified any named vendor’s retention, training or business data-protection terms for this page. Removing identifiers does not replace that review. Our ChatGPT overview provides broader product context.

Can ChatGPT review an offer or contract?

It can draft a summary and a list of questions, but the managing broker or attorney decides how to interpret and use the offer. Our offer-summary results do not identify which error types occurred, so do not assume any particular clause or deadline was handled correctly.

Use ChatGPT or Claude as a reading aid. Claude Pro includes Opus; that does not make a consumer-app session equivalent to our API test. This prompt is adapted from the tested offer-summary task:

You support a US listing agent. Draft an offer summary for review.
Include:
- A key-terms table.
- Every contingency deadline computed from the acceptance date,
  with the day-counting assumption stated.
- The 3–5 biggest seller risks in plain English.
- Questions for the buyer's agent.
Do not give legal conclusions. Flag missing or ambiguous terms.
End with what the seller's attorney or managing broker must review.
Offer:
[APPROVED, DE-IDENTIFIED OFFER TERMS AND EXPECTED ACCEPTANCE DATE]

Compare the summary with the original offer clause by clause. Recompute deadlines using the applicable contract language and have the reviewer confirm the counting assumptions. Keep unresolved questions visible rather than letting the assistant choose an interpretation silently.

Limits: what requires professional review?

This guide is not legal, tax or financial advice.

• Have a managing broker or attorney in your jurisdiction check offer terms, contingencies, deadlines and any summary intended for a client.
• Have the responsible broker review listing copy, captions and advertising distribution against applicable rules, including HUD guidance and NAR Article 12 where relevant.
• Review vendor terms, brokerage policy and professional confidentiality requirements before entering transaction data.
• Outside the US, obtain local guidance on advertising and professional obligations.

No prompt guarantees compliant copy or a correct contract summary.

FAQ