Perplexity releases pplx-embed-v2-late: 18B ColBERT distilled into 9B and 0.6B retrievers
beirmug · x · 2026-10-08
Perplexity officially released pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and page screenshots in a shared embedding space for cross-model querying. Per team member @bowangbo, they trained an 18B ColBERT teacher and distilled it into 9B and 0.6B versions (the 0.6B includes a vision tower). Sharing token embedding space enables 9B offline indexing with 0.6B online search. Both hit frontier performance and are on Hugging Face; a Dense variant is coming.
Related event: Perplexity Open-Sources Multimodal Embedding Models pplx-embed-v2-late(38 posts)→
More from Models
- ChatGPT Work mode vs Codex: same quota, far more tasks done per 5-hour window — sasik520 · 2026-10-08
- Epoch's InnovationEval: AI agents still far from producing real research innovations — Afinetheorem · 2026-10-08
- 113 decision models in 3 weeks: 70 built on Qwen, sub-cent per call — jonathanmalkin · 2026-10-08
- User reports Haiku 5.5 is a major workflow upgrade in screenshot post — Sorcerer12345 · 2026-10-08
- OpenRouter launches Decision Model Rankings, with typesafeai leading all categories — gaganghotra_ · 2026-10-08
- OpenAI launches Intelligent UI: ChatGPT now answers with fully interactive interfaces — gdb · 2026-10-08