Exa 3.0 Search API Outperforms Perplexity, Brave, and Parallel Across Eight Benchmarks

Exa has released Exa 3.0, an upgrade to its web search API built for AI agents. The Exa 3.0 search API introduces a larger embedding model and a lower-cost search tier called Exa Lite. Dynamic Highlights is now generally available.

Exa claims its API leads Perplexity, Brave, and Parallel across eight benchmarks. Because Exa ran those benchmarks internally, the results should be treated as self-reported. For developers, the most tangible change is pricing: Lite costs $4 per 1,000 requests.

TL;DR

• Scale: Exa indexes more than 2 trillion URLs and serves 100 billion pages. Embedding model size: not disclosed.
• Deployment: Hosted API only. Exa lists configurable search latency of 450 ms to 1 s. No self-hosted option is available.
• Performance: Exa 3.0 Auto ranks first on all eight Exa-run benchmarks against Perplexity, Brave, and Parallel.
• Strongest result: 91.8% on BrowseComp, plus the widest margin on ATLAS (42.0% vs. 37.2% for Perplexity).
• Weakest result: ATLAS at 42.0% is its lowest percentage score, and its BrowseComp lead over Perplexity is just 0.3 points.

via MarkTechPost