Exa AI Research Blog
Discover the latest in AI research and semantic search technology on the Exa blog. Learn how our neural network search engine provides high-quality web data for AI applications
TL;DR
- Exa 2.1 offers significant quality improvements across Exa Fast, Exa Auto, and Exa Deep search API endpoints.
- The Exa MCP quality has been dramatically improved with a new deep search tool.
- Exa Fast is now the most accurate search API with latency under 500ms.
- Exa Deep is positioned as the highest accuracy search API, utilizing multiple agentic searches for optimal results.
- Exa has built its own independent document index and retrieval algorithm, allowing for faster performance than APIs relying on Google Search.
- Evaluation methodologies for fast and MCP settings are detailed, using GPT-5 mini and gpt-4o-mini for grading.
- The company emphasizes the challenge and R&D involved in building a search engine from scratch, including semantic+lexical databases and large-scale crawling infrastructure.
- Future updates, such as Exa 2.2, are planned with continued scaling of techniques and infrastructure.