Grok Bot For Real Estate + Which AI Models To Pair With It: Live GoldieBench Pages And Landing Scores For Property Sites

Grok Bot for real estate does not let you choose its model, so the benchmark question for an agency is different: which models should build the property pages and listing sites that sit around the Bot in your Agent OS.

I read the live GoldieBench pages on 8 November 2026 and pulled the scores that matter for that job.

Grok Bot facts come from xAI's FAQ, files and results guide and computer and apps guide; every score names the GoldieBench page it came from.

This covers model choice for workflow, not legal advice, so listing copy and contracts still go past your broker.

Want help choosing for your own office? Book a free SEO strategy session.

Grok Bot picks its own model, so no score here is a Grok Bot score

The FAQ says Grok Bot picks the model for each task and there is no model picker.

xAI does not say which model runs a given request.

So the Grok rows below describe Grok on GoldieBench, a separate build test, and you should not read them as a measure of the Bot.

What the bench does help with is the rest of your stack: the model you use to build a listing microsite, a buyer comparison page or an open-house landing page.

What GoldieBench actually tests

The methodology page says every model gets the same fixed prompt, one shot, and must return one self-contained HTML file.

Scores are Julian's own 0 to 10 numbers, and the ranking is the average across the tasks a model was scored on.

That is a build test.

It does not test reading enquiries, checking calendars or getting property facts right, so it tells you nothing about accuracy on a listing.

The top of the board on 8 November

Every row is from the GoldieBench leaderboard.

LongCat-2.0 shows 8.12 but is marked provisional on four tasks, so it is not in the ranked list.

The Pages tab is the closest match to property pages

A property comparison page or a listing microsite is a single web page, so the Pages category is the nearest fit.

It has three tasks, which is a small sample.

Read from the Pages tab of the leaderboard.

The spread across the top eight is a third of a point, so treat these as one tier rather than a strict order.

The Landing task: a stand-in for an open-house page

The Landing task page asks for a modern marketing landing page in one shot.

It lists 26 models tested, 21 of 26 scored and a task average of 8.11 out of 10.

Four models share the top score of 9.0.

Grok's verdict on that page describes a clean nav, a gradient headline and two buttons built from one sentence.

For an agent, that means a tidy open-house or new-development page is within reach of several models, and the choice can come down to which one you already use.

The Visuals tab, for listing graphics and hero sections

If your microsite needs animated hero sections or rich graphics, the Visuals category is the next place to look.

It has nine tasks on the leaderboard's Visuals tab.

These tasks are things like particle effects and scenery, not property photos.

Never let a model invent or alter what a home looks like; use the real photos from the listing.

Grok on its own model page and deep dive

The Grok model page lists its three Pages builds: Landing 9.0, Webos 9.0 and Aipbpromo 7.0.

The Grok deep dive, marked "Data refreshed 2026-10-27", gives Pages 8.3 for Grok against a field average of 7.7.

Its lowest Page build, Aipbpromo, carries a short verdict calling it clean but plain.

Where the bench disagrees with itself

The methodology page still names Grok among models with zero scored tasks.

The model page carries a trade-off line saying none of its demos has a curated verdict yet.

Yet the leaderboard shows Grok seventh at 8.09 with 43 of 47 scored, and the deep dive says eighth of 26 ranked models.

Quote the dated leaderboard row, and check again before you repeat a rank.

A pairing for a small agency

1. Grok Bot runs the desk: it reads enquiries at its own address, drafts replies that wait for your Send, and holds viewing slots in Google Calendar.
2. For a buyer comparison, ask Grok Bot for a Page; the files and results guide says Pages hold tables, charts and filters and live in the Bot's Library.
3. For a public listing microsite, build it in your Agent OS with a model from the top Pages tier, then have a person check every fact.
4. Keep the source notes in one Drive folder so both sides work from the same facts.

Because Enterprise teams cannot use Pages yet and other accounts get them in stages, the Agent OS build route doubles as the fallback.

Run your own bake-off before you commit

1. Write one prompt for the page you actually need, such as an open-house page for a single property.
2. Give it to two or three models from the top Pages tier inside your Agent OS, one shot each.
3. Feed every model the same listing notes and real photos.
4. Check each page for facts that are not in your notes, then score it yourself.
5. Keep the winner's prompt as a skill so the next listing starts from it.

That mirrors how the bench works: same prompt, one attempt, judged on the result.

Members of the AI Profit Boardroom share their own bake-off prompts, and my team can review your agency pages on a free call.

What the scores cannot tell an agent

No GoldieBench task checks whether square footage, room counts or school names are right.

None tests local advertising rules for listing copy.

And none tests email or calendar work, which is most of what the Bot does all day.

Use the scores to pick a page builder; use review and your broker for everything else.

For the same question asked about a sales Bot, see Grok Bot for sales on GoldieBench.

Also On Our Network

• Agent Operatorsthe ten-step listing desk build with aliases, Ask first rules and a tested routine
• AI Income Deskwhether a small agency should use it, with five office examples step by step
• AI Tool Verdictthe eight-criterion agency scorecard and a 6.6/10 verdict from the docs
• agentos.guidea ten-minute property desk quick start with five copy-paste prompts
• aiprofitboardroom.comthe complete Grok Bot for Real Estate guide on the AI Profit Boardroom blog

Pick a page builder from the top Pages tier, let the Bot run the desk, and join the AI Profit Boardroom to wire both into Grok Bot for real estate.