Perplexity's new Search API gives developers access to an index of hundreds of billions of webpages with source corroboration built in.
Perplexity launched its Search API on September 26, 2025, giving developers access to the same internet-scale infrastructure that powers its public answer engine. The API indexes hundreds of billions of webpages and includes the source corroboration capabilities that distinguish Perplexity from traditional search.
What Developers Get
The Search API provides programmatic access to Perplexity's retrieval infrastructure — the same system that fetches, ranks, and corroborates sources for answers in the consumer product. Key capabilities include:
- Internet-scale indexing with workflows optimized for content freshness and real-time accuracy
- Source corroboration built into retrieval, not bolted on afterward
- Multi-step query support for both single-step lookups and deep research workflows
- A Search SDK for rapid prototyping — Perplexity engineers report building product prototypes in under an hour
The company also released an open-source evaluation framework called search_evals, enabling developers and researchers to benchmark search APIs against competitors.
Performance and Pricing Claims
Perplexity says its API outperforms competitors across both single-step queries and deep research workflows at lower cost. The evaluation framework is designed to let developers verify those claims independently rather than relying on marketing benchmarks.
The API Platform now houses the developer console and documentation for both the Search API and the existing Sonar APIs, centralizing resources for teams building retrieval-augmented applications.
Integration Patterns
For developers building AI applications, the Search API addresses one of the most persistent engineering challenges: reliable, fresh retrieval at scale. Common integration patterns include:
RAG pipelines. Replace self-hosted crawlers and vector databases with Perplexity's managed retrieval for question-answering systems.
Agent tool calls. Give autonomous agents a search tool that returns corroborated results rather than raw web snippets.
Research automation. Build workflows that require multi-step information gathering with source attribution.
Content freshness monitoring. Track how indexed content changes over time for competitive intelligence or news aggregation.
Getting Started
Developers can access the API through Perplexity's developer console. The Search SDK supports rapid integration testing. Perplexity plans to showcase the API at its San Francisco API Day and London hackathon in the coming month.
Community forums provide ongoing support for implementation questions.
Why This Matters
Search infrastructure has been a bottleneck for AI application development. Teams either build custom crawlers and indexes — expensive and maintenance-heavy — or rely on general-purpose web search APIs that return unranked, uncorroborated results.
Perplexity is betting that the same retrieval quality that makes its consumer product trustworthy can become a platform primitive for the entire AI ecosystem. If the performance claims hold up under independent evaluation, the Search API could become a standard dependency for production AI applications — similar to how Stripe became the default payments API for web applications.
For developers evaluating retrieval options, the search_evals framework offers a structured way to compare Perplexity against alternatives using your own query sets. That transparency is itself a competitive differentiator in a market where benchmark claims are often impossible to verify.
Comments
Loading comments…