Perplexity launches Monitor API for continuous web tracking
Perplexity has released a new Monitor API that automatically searches the web on a schedule and sends alerts to AI agents only when meaningful changes are detected. The service uses a built‑in decision model to filter results for relevance, credibility and timeliness.

Key takeaways
- Perplexity's Monitor API automates scheduled web searches and pushes only relevant changes to AI agents.
- A built‑in decision model scores results for relevance, credibility, novelty and timeliness before sending alerts.
- Pricing starts at one dollar per thousand runs for web monitors and per thousand URLs for page monitors.
- Early internal benchmarks show the service returning more fresh results than several rival monitoring tools.
In this article
What the Monitor API does
According to The Decoder, Perplexity's Monitor API turns a one‑off search into an ongoing watchdog. Developers define a topic or a list of up to ten specific pages, and the API runs searches at intervals ranging from every fifteen minutes to once a week, or on a custom cron schedule. Each run resumes where the previous one stopped and removes duplicate findings. When the system decides that a change matters, it delivers a signed webhook payload with a confidence score.
How results are filtered
The API evaluates every candidate result against a goal expressed in natural language. For web searches it checks relevance, source credibility, novelty and timeliness; for page monitors it judges whether the modification matters to the stated goal. Only when all thresholds are satisfied does the service send an update. Developers can adjust each threshold independently and modify the goal without recreating the monitor.
Pricing and availability
Web monitors cost one dollar per one thousand successful runs, while page monitors cost one dollar per one thousand fetched URLs. The Monitor API is live now and can be accessed through the same key used for Perplexity's Search, Context and Agent APIs. Controls for pausing, resuming or manually triggering a monitor are provided both in the API and in an online console.
Competitive context
Perplexity ran internal tests against monitoring offerings from Firecrawl, Exa and Parallel across sixty real‑world topics. In the first eight hours the company reported delivering sixty‑three fresh, relevant pages versus twenty for Firecrawl, fifteen for Exa and ten each for the two Parallel tiers. Those figures are described as early internal results. The firm is positioning the Monitor API as part of a broader push to differentiate its developer platform as web search becomes a standard feature in mainstream chatbots.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



