Field noteOct 22, 20261 min read1 source

Exa Ranks #1 on the Artificial Analysis Search Index

In one minute

Exa said on October 19, 2026 that Exa 3.0 ranked first for quality in the Artificial Analysis Search Index.

What changed: Exa 3.0 now ranks #1 on the Artificial Analysis Search Index, using the fewest queries per task; the prior covered Parallel’s ranking on the same benchmark.

The company describes the index as an independent benchmark for search providers and says Exa 3.0 required the fewest search queries per task.

The release says agents using Exa can make 20% to 50% fewer search queries per task than competing providers while reaching higher accuracy. It also claims that DeepSeek Search needs 1.5 times as many queries as Exa. Exa 3.0 includes a larger embedding model, Dynamic Highlights that the company says use 95% fewer tokens than full-page text, and billions of added documents across products, local businesses, flights, events, sports, and weather.

Dynamic Highlights are generally available, according to Exa, and the company directs users to its API dashboard and documentation. The post does not reproduce the benchmark's scores, task set, provider configuration, pricing comparison, or methodology. Its performance and cost claims therefore cannot be independently established from this source alone.

Evidence / sources

  1. 1exa.ai/blog/artificial-analysis-exa-3-best-search-api
All Field Notes

Next reads

Read next

Open the graph
  1. 01topicAgent EconomicsExplore the agent-economics topic hub.
  2. 02topicModel RoutingExplore the model-routing topic hub.
  3. 03archiveField NotesOpen the latest editorial analysis.
  4. 04source ledgerSource LedgerInspect the public source evidence graph.
Who read this page?Machine requests, hidden until opened

Loading the privacy-safe route aggregate...

Open the JSON contract