Perplexity AI has released ‘search_evals’, an open source framework that lets developers benchmark search APIs, marking a direct challenge to Google’s dominance as it launches its massive new Search API.
Perplexity AI has taken aim at Google’s long-standing grip on search by releasing ‘search_evals’, an open source framework that allows developers to independently benchmark the quality and performance of search APIs. By putting this power directly in the hands of developers, Perplexity is breaking the closed-wall model of search infrastructure that has traditionally restricted transparency and innovation.
The framework launches alongside Perplexity’s new Search API, a comprehensive service that grants developers access to the same massive web index powering the company’s answer engine. With real-time updates across hundreds of billions of web pages and AI-optimised formatting, the Search API is being positioned as the most serious challenge yet to Google’s dominance in developer-facing search infrastructure.
“Legacy search engines have kept developers beholden to their interests, namely, favouring commercial intent traffic over helpful content,” said Beejoli Shah, Spokesperson, Perplexity AI. The company argues that by combining an open source evaluation tool with a commercial-scale API, it is democratising both access and accountability in a space historically tilted towards established players.
Technical innovations in the Search API include real-time indexing (making content searchable within seconds), sub-document precision (ranking passages instead of entire pages), hybrid retrieval (blending semantic and keyword search), and a citation system that surfaces contextual snippets with built-in attribution. Together, these features are designed to serve AI developers who require fresher, more targeted, and source-credited data.
The move comes at a pivotal moment, as Google faces mounting antitrust pressure and developers seek alternatives to increasingly restrictive APIs from traditional providers. If Perplexity’s open source approach succeeds, it could reshape not only who controls access to the world’s search data but also how AI-driven applications are built and validated in the years ahead.
