Typed JSON, Always

Strict coercion on every field: strings, integers, floats, booleans, dates, and arrays. Control each field to ensure no more ambiguous values slip through.

Plain JSON Schema

No CSS selectors, XPath, or complex DSLs. Simply define your field names, types, and examples to create a schema any developer can write in 30 seconds.

Static & Dynamic Keys

Bind a schema to a static key once and future requests require only a URL. Alternatively, attach a custom schema to every request using your dynamic API key, enabling higher flexibility.

Three Methods of Scraping

Scrape a single URL with /extract, fan-out to multiple URLs with /batch, or crawl recursively with /crawl using the same schema on the same API key.

Semantic Extraction

Runo extracts by meaning, not DOM position. Site redesigns and HTML changes won't ever break your pipeline.

Explicit Nulls

Unresolvable fields return null. Runo never quietly drops, never omits and never foolheartedly guesses a wrong response. You always know exactly what was and was not found.

Field Hints

In the case of a specific request, add a hint string to any field to have it injected directly into Runo's extraction instructions, ensuring you get exactly what you need.

Auto JS Rendering

Transparent fetch-first strategy. Auto-escalates to Playwright headless when JS detection or 403 triggers. "render_mode" in the response tells you exactly which path ran.

Bot Bypass

CAPTCHA solving and residential proxy access included with Pro & Scale plan. Runo automatically detects and escalates the right bot detection and bypassing systems to ensure you get the data you called for, no matter the website or their defense.

Define your schema

List the fields you want. Name, type, example and add a hint if necessary.

Call the API

POST to /extract, /batch, or /crawl with your key and schema and Runo handles the rest.

Runo fetches and renders

Plain HTTP first. Auto-escalates to headless if JS-gating or empty response detected.

Get typed JSON back

Runo extracts and coerces every field. Clean validated JSON. Unresolvable fields return null.

Yes. Model selection is automatic and included in the request cost.

Runo caps crawl jobs at 25% of your remaining monthly quota per job, so a single runaway crawl can't exhaust your account. Unused reserved pages are refunded immediately when the crawl finishes or is cancelled. If you need to go beyond the cap, pass "allow_large_crawl": true at your own discretion.

Yes. Every response includes an X-Job-ID header. Send DELETE /v1/jobs/{job_id} with the same API key to cancel. The job stops at the next safe checkpoint, returns whatever it completed, and refunds unused quota including overage credits.

Runo's free tier gives you access to 500 requests per month at a 1 request/min rate, along with the same extraction quality as paid tiers. E.g. JS rendering, stealth mode, and all extraction features are included. The only thing not included in the free plan are the automatic enhanced bot bypassing features and image parsing.

You get a 429 QUOTA_EXCEEDED error. On Pro and Scale, you can purchase prepaid overage credits in advance and the API draws from that balance automatically. There's no surprise bill after the fact, if you have no credits, requests stop. Credits don't expire and are restored if you cancel and re-subscribe.

Competitors price by credits, where a single large or JS rendered request can cost 5-75 credits. Runo charges flat per request. At the Scale tier, Runo works out to $0.90 per 1,000 effective requests vs. around $6 for the nearest Firecrawl equivalent.

P50 latency is around 10-14s for the happy path. Pages requiring headless rendering add a few seconds. The worst case, a full headless render on a blocked site, can reach around 45s at P50.

You can try add a "hint" to the field to direct the extraction. Check that your "type" matches what's on the page (e.g. a price shown as "$29" needs float, not string); try render_js: "always" if the page is JS-heavy. If nulls persist across retries, the data may be in an image. The process_images: true option (Scale only) runs a vision pass over scored image candidates to fill remaining nulls.

None. One 'POST' request with a JSON body, one JSON response.

Yes, we consider refund requests on a case-by-case basis. If you think you're owed a refund, email [email protected] with your request and reason, your payment receipt or order confirmation, and a description of the situation. We'll review it promptly and let you know whether a refund is possible.