Radar Labs · free dataset · look up a company · method

Run the job + salary API in 60 seconds

One call returns rows like this — 8,039 postings across 266 companies already carry a published annual range; the median is $200,000:

{"company":"anthropic","ats":"ashby","title":"Software Engineer, Inference",
 "location":"San Francisco, CA","remote":false,"salaryLow":280000,"salaryHigh":420000,
 "salaryCurrency":"USD","salaryUsdLow":280000,"salaryUsdHigh":420000,
 "postedAt":"2026-10-21T18:04:00+00:00","url":"https://job-boards.greenhouse.io/..."}

1. Get a token

Free Apify account → Settings → Integrations → API token. The dataset CSV needs no account at all; the API does.

2. Call it

curl -s -X POST \
  "https://api.apify.com/v2/acts/zqBGjpzG8z3UZxN51/runs?token=$APIFY_TOKEN&waitForFinish=300" \
  -H "Content-Type: application/json" \
  -d '{"useVerifiedPool": true, "salaryOnly": true, "maxJobs": 50, "maxPerCompany": 2, "maxWorkers": 6}'

Prefer the console? https://apify.com/contributory_garnet_hyz/ats-job-feed

3. What it costs

$0.0005 per run start plus $0.0015 per row returned. Rows you do not get are not charged, and postings without a salary are never returned when salaryOnly is on — so filtering is free.

Rows	You pay	What that buys
50	$0.0755	fast: descriptions off, no salary parsing
200	$0.3005	reads job descriptions to recover pay ranges
1,000	$1.50	reads job descriptions to recover pay ranges

Platform compute is billed separately and is roughly $0.006 per full-pool run with descriptions on, $0.0008 with them off.

4. Why the numbers can be trusted

Salaries are normalised, not guessed. Structured fields first (Ashby summaryComponents, Lever salaryRange); only then a text window around pay-cue phrases.
Implausible values are dropped, not clamped. Anything outside $5k–$5M a year after normalisation is removed entirely, so the tail is missing rather than wrong.
Currency uses a daily rate and the original local range is kept in salaryLocal*, so you can re-derive it yourself.
Ranges are ranges. These are published bands, not what anyone was actually paid — 43% of the 18,668 postings we read publish nothing at all.

5. If you would rather not pay

The same rows are a free CSV, CC BY 4.0: 8,039 rows, 1.7 MB · method and distributions · mirror on GitHub. The paid version exists for the things a CSV cannot do: filters, incremental runs, and being a stable endpoint.