Perplexity Launches Search API Claiming SOTA Performance on SimpleQA and SEAL Benchmarks
Key Facts
- Perplexity has released a Search API that it says achieves state-of-the-art (SOTA) results on all major industry benchmarks, including SimpleQA and SEAL.
- The API provides developers direct access to Perplexity’s proprietary index containing over 200 billion URLs, refreshed in real time.
- Developers can integrate real-time web information with citations directly into their own applications and AI agents.
- The announcement was made via Perplexity’s official X account, confirming it as a first-party product launch.
- Multiple competing search APIs, including those from Linkup, Brave, Tavily and Exa, have also recently claimed SOTA scores on SimpleQA, highlighting intense competition in AI-grounded search.
Perplexity AI has unveiled its new Search API, positioning the product as the current leader across key industry benchmarks used to evaluate retrieval and factuality in AI systems. According to the company’s official announcement, the API delivers state-of-the-art performance on benchmarks including SimpleQA and SEAL while giving external developers access to the same massive, real-time web index that powers Perplexity’s popular AI search engine.
The launch comes at a time of rapid innovation in the AI search and retrieval-augmented generation (RAG) space. Several companies have recently reported record scores on SimpleQA, a benchmark designed to test an AI system’s ability to answer short, fact-based questions using web search. Perplexity’s entry raises the bar further by claiming leadership across “all industry benchmarks,” including the more complex SEAL evaluation suite.
Technical Capabilities and Index Scale
Perplexity’s Search API offers developers direct access to an index of more than 200 billion URLs that is refreshed in real time. This scale and freshness are critical for building reliable AI agents and applications that require up-to-date information rather than relying on static training data.
The company emphasizes that the API returns results with citations, allowing developers to build transparent, verifiable applications. This citation feature addresses one of the core challenges in current large language model deployments: reducing hallucinations by grounding responses in verifiable web sources.
According to the official announcement, the Search API enables developers to “call real-time information with citations inside your own apps and agents.” The product is positioned as a straightforward way to add Perplexity-grade search capabilities to third-party products without having to build and maintain a custom web index.
Competitive Landscape and Benchmark Claims
The AI search sector has become increasingly competitive, with multiple players claiming state-of-the-art results on SimpleQA in recent months. Linkup reported a 91.0% F-Score, Brave’s AI Grounding service claimed 94.1%, and Tavily announced 93.3% accuracy on the same benchmark. Exa has also positioned its neural search engine as a top performer for LLM use cases.
Perplexity’s claim of SOTA performance across “all industry benchmarks including SimpleQA and SEAL” is notable because SEAL is generally considered a more challenging evaluation that tests deeper reasoning and multi-step retrieval capabilities. The company’s first-party announcement has been verified as an official product launch, though independent third-party verification of the exact benchmark numbers was not available at the time of publication.
This wave of benchmark claims reflects the strategic importance of retrieval systems in the broader AI ecosystem. As large language models grow more capable, the quality of their grounding data has become a primary differentiator. Companies are racing to build specialized indices and ranking systems optimized specifically for AI agents rather than traditional web search.
Implications for Developers and AI Agents
The availability of Perplexity’s Search API could significantly lower the barrier for developers building AI-powered applications that require fresh, factual information. Rather than depending on general-purpose search APIs or attempting to crawl and index the web themselves, developers can now tap into an index explicitly designed to support large language models.
The real-time refresh capability is particularly valuable for applications in finance, news, research, and customer support, where information changes rapidly. By providing structured results with citations, the API also helps developers meet growing demands for transparency and auditability in AI systems.
Enterprise adoption of retrieval-augmented systems has accelerated as organizations seek to deploy AI agents that can reliably interact with current data. Perplexity’s move follows similar launches from other specialized search providers, suggesting the market is maturing toward standardized interfaces for AI-grade search.
What’s Next
Perplexity has not yet disclosed detailed pricing, rate limits, or specific model versions associated with the Search API. The company is expected to release additional technical documentation and integration guides in the coming weeks.
The competitive benchmark race on SimpleQA and SEAL is likely to continue, with further claims and counter-claims expected as companies refine their retrieval architectures. Independent leaderboards and standardized evaluation frameworks may emerge to help developers compare offerings more objectively.
For the broader AI industry, the proliferation of specialized search APIs signals a shift toward modular architectures where frontier models can be paired with best-in-class retrieval systems. This composable approach could accelerate innovation while allowing companies to focus on their core strengths, whether in reasoning, search, or application-layer experiences.
The launch underscores Perplexity’s evolution from a consumer-facing AI search engine to a full-stack AI platform provider. By opening its core index to developers, the company is betting that widespread adoption of its search infrastructure will strengthen its position in an increasingly crowded market.