Perplexity releases pplx-embed-v2-late: open late-interaction embeddings for text, image and pages
antoine_chaffin · x · 2026-10-08
Perplexity has released pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages in a shared embedding space for cross-modal querying.
Both models reportedly achieve frontier retrieval performance and are publicly available on Hugging Face. Community response highlights late interaction as the emerging frontier retrieval paradigm, with the smaller model still reaching the Pareto frontier against strong competition.
More from Models
- How Long Before Anthropic Ships a Decision Model Like Jev? — tristanbob · 2026-10-08
- Orgs now switch between 12 models a year; blended token cost fell to $0.14 — Tiancaixinxin · 2026-10-08
- Liquid AI open-sources d1 decision models: 3B model answers in one forward pass, 8ms on RTX 4090 — huggingface · 2026-10-08
- Hands-on: OpenAI's new Decisions API edges out Jev in accuracy and image classification — cantrell · 2026-10-08
- New Haiku matches last year's frontier, meaning Claude Code-class capability is heading toward 'too cheap to meter' — johnvmcdonnell · 2026-10-08
- Haiku 5.5 already burns more tokens at medium than Opus 5.5 does — haider1 · 2026-10-08