Perplexity open-sources pplx-embed-v2 contextual embedding models in 9B and 0.6B sizes
antoine_chaffin · x · 2026-11-03
Perplexity has released contextual embedding models pplx-embed-v2-context on Hugging Face in two sizes: 9B and 0.6B. They are fine-tuned from pplx-embed-v2-late-9b on a Qwen3.5 architecture under MIT license. The author notes the results are remarkably strong given they come almost entirely from zero-shot transfer of the ColBERT backbone plus contextual text training. An accompanying paper (arXiv:2610.38854) is packed with details.
More from Models
- Only recent models seem to know "weight surgery" — are labs building Bayesian weight priors? — Tkaraletsos · 2026-11-03
- 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