UltraBERT Fits 32K Tokens in One Vector. RIP Chunking?
Dylan Couzon, Anush
October 19, 2026
Qdrant 1.20 - Full-Text Search Improvements
Andrey Vasnetsov
October 22, 2026
- Products
- Solutions
- Developers
- Resources
- Company
Qdrant Blog
Scalable PDF Retrieval with Cohere Embed 5 and Qdrant Asymmetric Quantization
Use Cohere embeddings for visual PDF retrieval in Qdrant. Compare ViDoRe results and combine small vectors in RAM with full vectors on disk.
Evgeniya Sukhodolskaya
October 23, 2026
How Poliscope Made All of Germany's Local Politics Searchable on One Server with Qdrant
How Poliscope indexed about 4,800 German council portals into one self-hosted Qdrant collection of 130M points: hybrid search, tenant and prefix filtering, and TurboQuant on one bare-metal server, taking mean search time from 20 seconds to 0.87 seconds.
Daniel Azoulai
October 15, 2026
Jevjitsu: Or, How We Tried Generalized Classifiers on Everything
We applied the Jev generalized classifier to reranking, query understanding, near-duplicate removal, and semantic chunking.
Andrei Cristea & Chadha Sridi
October 08, 2026
How Small Can Google's New EmbeddingGemma 2 Get?
Google's EmbeddingGemma 2 in Qdrant: 99% of retrieval quality in 30x less vector RAM, or 94.5% in 77x less with rescoring.
Qdrant Team
October 06, 2026
Constella Preview: Swap Query Models Without Re-Embedding
Swap query models without re-embedding documents. Explore Constella Zero, Nano, and Stella, with BEIR-15 results and CPU benchmarks.
Dylan Couzon
September 29, 2026
Oxidizing Cross-Encoders
Learnings from writing a cross-encoder inference library in Rust and comparing it with existing Python packages.
Clelia Bertelli
September 25, 2026
How Lucy Built a Source-of-Truth Retrieval Layer for 2.8M+ SEC and DART Filings with Qdrant
How Lucy built a retrieval layer over 2.8M+ SEC and DART filings on Qdrant: four-lane hybrid search, payload filtering, and TurboQuant, reaching 94.7% Recall@10 on FinanceBench.
Daniel Azoulai
September 21, 2026
How Himalayas Built Semantic Job Matching for 100,000+ Remote Listings with Qdrant
How Himalayas moved remote job matching from keyword search to hybrid retrieval on Qdrant Cloud: 25M+ requests a month, around 30 ms median time in Qdrant, and a 19-point recall gain.
Daniel Azoulai
September 18, 2026
SHIFTing Languages in Multilingual RAG
Fix multilingual embedding language bias with SHIFT: shift document and query vectors to one pivot language and store them in a single Qdrant index.
Evgeniya Sukhodolskaya
September 16, 2026
Get Started with Qdrant Free
Get Started