Perplexity open-sources its 0.6B and 9B multimodal embedding models

KuCoinFlash

Release Time: 2026/10/08 12:18:30

Summary

Perplexity has open-sourced two multimodal embedding models—pplx-embed-v2-late with 0.6B and 9B parameters. The models support text, image, and PDF retrieval using a 128D vector per token. Developers can use the 9B model for indexing and the 0.6B model for queries. Trained by an 18B teacher model, the 0.6B model achieved a score of 62.3% on ViDoRe v3 tests, increasing to 63.5% when using the 9B model for indexing. Model weights are available on Hugging Face under the MIT license. This update introduces new token listings and marks a key token launch announcement.

ME AI announces that Perplexity has open-sourced two multimodal embedding models, pplx-embed-v2-late, with sizes of 0.6B and 9B. The vectors of both models are mutually compatible, allowing developers to build a database using the 9B model while using the 0.6B model for everyday searches, eliminating the need to run the larger model for every query. The new models also support text, image, and PDF page retrieval. Traditional embedding models typically compress a segment of content into a single vector, often losing detail. These new models preserve a 128-dimensional vector for each token, enabling search terms to match relevant parts of the document precisely. When processing PDFs, PPTs, and scanned documents, the models can directly convert page images into vectors without requiring OCR text extraction, while retaining charts, tables, and layout information. Both models were trained from the same 18B teacher model and share a unified vector space. In the ViDoRe v3 image retrieval benchmark, using the 0.6B model for both database building and querying achieved a score of 62.3%; switching to the 9B model for database building and the 0.6B model for querying improved the score to 63.5%, while using the full 9B model for both tasks achieved a score of 65.2%. The model weights are now available on Hugging Face under the MIT license. (Source: BlockBeats)

Source:Show original

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.