Perplexity launches Context API for multi-page web extraction
Perplexity's Context API extracts content from supplied URLs and lets developers choose between cached and live pages for retrieval workflows.
Perplexity launched its Context API, an extraction service that takes web addresses and returns page content for applications such as agents and retrieval-augmented generation systems. The company says the launch interface can read as many as 10 URLs in one call and return Markdown, HTML, page links or passages selected for relevance to a query.
Developers can control how recently a page must have been fetched. Perplexity’s documentation sets max_age_seconds to 172,800 by default, allowing a stored copy fetched within the previous two days. If no sufficiently recent copy is available, the service fetches the page live. Setting the value to zero forces a live fetch. The cache_only option prevents one and returns a no_content status when no qualifying stored copy exists. Response metadata identifies whether the result came from a cached or live page and records its fetch time.
Requests go to the documented POST /v1/context endpoint with the target URLs and requested output formats. Perplexity says it fetches the URLs in parallel, while the response preserves request order and reports a separate status for each URL. That structure allows an application to use successful results when another page fails or times out. For highlights, callers provide one query and can set a token limit per URL; the company describes those passages as selections from the source page rather than generated summaries.
The launch limit comes with a qualification. Although the announcement described as many as 10 URLs per call, the current documentation lists tier-dependent limits of 10, 15 or 25 URLs, with a process to request a limit of as many as 50.
Context API handles a different step from Perplexity’s other developer interfaces. Its Search API begins with a query and returns ranked web results as structured JSON. Context API starts with URLs the caller already has and retrieves their content. Perplexity’s separately launched multi-model Orchestrator API, by comparison, accepts prompts and provides direct model access with optional per-request model routing.
Perplexity lists the Context API price at $1 per 1,000 billable URLs, counting results marked ok or not_modified and excluding failed URLs. The company also published evaluation scores of 92.3% usable results, 87.9% fact recall and 80.3% content precision. The available issuer materials do not provide enough reproducibility detail to validate those figures independently.
More news

Perplexity launches Monitor API for recurring web alerts

Google launches Gemini 4 Neon across API, business and paid consumer tiers

Perplexity launches Orchestrator API for multi-model inference
