Perplexity's 9B contextual embedding sets SOTA, scores 45.5 on context-bench vs 31.1 for voyage
antoine_chaffin · x · 2026-11-03
Perplexity's antoinechaffin shared results for pplx-embed-v2-context: the 9B model hits 45.5 Answer@10 on turbopuffer's context-bench, well above voyage-context-4's 31.1.
It also claims SOTA on ConTEB (without training on its data), context-bench, and the newly introduced Q2D-Chunk benchmark, and leads on evidence retrieval and document disambiguation — finding the right document among near-identical ones while surfacing both answer and supporting context. A 0.6B sibling is strong, cheaper to serve, and can even query indexes built with the 9B model.
More from Models
- Victor Taelin floats labs replacing backprop with models editing their own weights at scale — arthurcolle · 2026-11-03
- Custom fine-tunes now beat frontier models at ~1/20th the cost, says investor — bindureddy · 2026-11-03
- OpenAI updates Model Spec: intent over literal instruction-following — sven_ai · 2026-11-03
- Zyphra's ZHA brings subquadratic attention via hierarchical tree-structured indexer — yacineMTB · 2026-11-03
- LiquidAI's LFM3-1.7B tops HF trending with 1k likes and 42.4k downloads in 5 days — maximelabonne · 2026-11-03
- 'Gemini is your friend and alignment is complicated': another eyebrow-raising model reply — vedantmisra · 2026-11-03