Perplexity releases contextual embedding model with new SOTA training approach

antoine_chaffin · x · 2026-10-01

Perplexity researcher Antoine Chaffin announced a preview release of their contextual embedding model along with a new training approach. Contextual models address the chunking problem in retrieval by encoding passages with query context, and the new method pushed SOTA results including on a fresh benchmark from turbopuffer. Technical details are discussed in the accompanying blog post.

Related event: Perplexity releases pplx-embed-v2-context-9b-preview, tops ConTEB embedding benchmark(5 posts)→

Original post →

More from Research

Research channel →