Benchmark: search homes on Zillow with an AI agent
Wield's recipe found homes on Zillow in 9 of 9 runs, with a median of 0.9 s and no model cost. A Claude browser agent passed 0 of 9 runs, with a median of 14.8 s and $0.01 per run. The benchmark flagged all 9 agent runs as blocked by the site. The low cost is not a sign of efficiency, because blocked runs end early.
Zillow Search US homes for rent or for sale in a city
Read-only, no loginPasses the daily check
Parameters (5)
| Metric | Wield recipe | Browser Use + Claude Sonnet 5.5 |
|---|---|---|
| Success | 100% | 0% |
| Median time | 896 ms | 15 s |
| Cost per run | $0 | $0.013 |
9 runs each, 7 November 2026. How we measure
The task
Each side ran the same three searches. A run passes when it returns at least five homes with a name or address and a price.
| Query | Parameters |
|---|---|
| q1 | city = Austin, TX, listing_type = rent, max_price = 2500 |
| q2 | city = Denver, CO, listing_type = rent, max_price = 3000 |
| q3 | city = Seattle, WA, listing_type = buy, max_price = 800000 |
Results
| Wield recipe | Browser agent | |
|---|---|---|
| Runs passing the check | 9 of 9 | 0 of 9 |
| Median time per run | 896 ms | 14.8 s (14,815 ms) |
| Model cost per run | $0 | $0.01 |
| Requests per run (median) | 1 | 17 (to the site: 2) |
| Blocked by the site | 0 runs | 9 of 9 runs |
The browser agent took a median of 2 steps per run, with 2,766 input tokens and 407 output tokens. Its requests to zillow.com returned HTTP 403 on the homepage in every run. Wield had no failed runs. Its fastest run took 0.6 s and its slowest 5.8 s.
Why the results differ
The browser agent stopped at the first page. Zillow's protection returned 403 on the homepage, so the agent could not reach the search results. Our recipe notes name PerimeterX (HUMAN) as the protection on zillow.com.
Wield sends one search request from a tab in the user's own browser. It runs at human pace and uses no model.
Caveats
- Small sample. The benchmark is still running. It is paced at about 30 site requests per hour, so counts are low on some sides.
- Cost is misleading here. The agent's $0.01 per run comes from runs that failed early. Do not read it as an efficient result.
- Human pace only. Wield does not solve bot protection checks. If Zillow shows a check, Wield stops with
site_challengeand the user solves it in their own tab.
Terms of Zillow
Zillow's terms prohibit automated queries and scraping. The recipe makes one search page at a time that the user starts, from the user's own browser, with 1 request. Check the site's terms before you use any automation.
FAQ
Does Zillow block browser agents?
In our runs, yes. A Claude browser agent in a fresh headless Chromium was flagged as blocked in all 9 runs. Wield's recipe passed 9 of 9. The result comes from one agent setup on one date.
How fast is the Wield recipe for search homes on Zillow?
In our benchmark as of 7 November 2026, the median run took 0.9 s. The fastest run took 0.6 s and the slowest 5.8 s.
Can I rerun this benchmark?
The method and per-task rows are on the benchmark page. The raw logs hold destinations, timing, tokens and check results, not page contents.