Top all-time · #3379
Vespa Blog
Company blog · blog.vespa.aidataaisystems
Vespa Blog is the company blog at blog.vespa.ai, writing about data, ai and systems. Since 2017, 23 different Hacker News readers have linked it 72 times, most often “Billion-scale vector search using binary-coded representations”. It is still writing: the last post was on 2026-10-23, and you can follow it by feed.
Too few sampled comments (14) to say how often HN recommends it.?
Why HN links it
HN readers link the Vespa blog as a technical resource on vector search, hybrid sparse and dense retrieval, billion-scale kNN, multi-vector indexing and multi-stage ranking. Most comments simply cite posts as support.
Summarised by Claude from a sample of 14 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
The AI Search Platform
Most-linked posts?
- Billion-scale vector search using binary-coded representations5 people · 6 mentions · first linked 2021-12-04
- /vespa-hybrid-billion-scale-vector-search/3 people · 4 mentions · first linked 2022-09-06
- Building a billion-scale vector search solution without breaking the bank2 people · 2 mentions · first linked 2022-10-07
- Perplexity builds AI Search at scale on Vespa.ai2 people · 2 mentions · first linked 2025-04-21
- Scalable evaluation of ONNX models with open source Vespa2 people · 2 mentions · first linked 2018-06-25
- An open source scalable shopping back end w. navigation, search, recommendation2 people · 2 mentions · first linked 2019-05-03
- Introducing TensorFlow support in Vespa2 people · 2 mentions · first linked 2018-03-14
- Vespa.ai is spinning out of Yahoo as a separate company2 people · 2 mentions · first linked 2023-10-04
What Hacker News says?
Yes, Vespa.ai can index billion scale vector datasets https://blog.vespa.ai/billion-scale-knn/ Vespa also allows expressing hybrid sparse and dense retrieval (WAND for sparse, ANN via HNSW for dense). It's also easy to express multi stage retrieval ranking phases, as vector search alone is not achieving state-of-the-art ranking results, see https://blog.vespa.ai/pretrained-transformer-language-models...
Comment linking /pretrained-transformer-language-models-for-search-part-4/
on “Find anything fast with Google's vector search technology”Submitted Open Sourcing Vespa, Yahoo’s Big Data Processing and Serving Engine as a story · 66 points
HN discussionNot OP but I am working on a search engine with vector ranking. Why do you say that vector search would help with narrow queries? In my experience, semantic search helps broaden the query to search for adjacent ideas without exact term marches. Hybrid approaches that use vector search for broad matches and rerank using BM25 could be what you’re looking for. See https://blog.vespa.ai/efficient-open-domain-question-an…
Comment linking /efficient-open-domain-question-answering-on-vespa/
on “Teclis – Non-commercial web search”Submitted Building a Billion-scale vector search as a story · 22 points
HN discussionHybrid search with vector similarity and filtering I think has mostly been solved by Vespa and not even recently. https://blog.vespa.ai/vespa-hybrid-billion-scale-vector-sear...
Comment linking /vespa-hybrid-billion-scale-vector-search/
on “Scaling HNSWs”Submitted How we made neural nets inference 20x faster with Vespa as a story · 22 points
HN discussionI migrated from ES and for me, I do not agree about the feeling that it doesn't feel modern. The Middleware logic container and Live reconfiguration it is mind blowing. About those two things: - Nested (For my use cases, this is a problem I do not have. For more complex cases, there is parent-child https://blog.vespa.ai/post/174589826190/parent-child-in-vesp... ) - Exact match ( use the exact match https://docs.vesp…
Comment linking /post/174589826190/parent-child-in-vespa
on “Elasticsearch from the Bottom Up (2013)”Submitted Billion-scale vector search using binary-coded representations as a story · 10 points
HN discussionShameless plug from someone who want this project to flourish. Check https://vespa.ai as an alternative to Elasticsearch. Migrating from a ES to it, I got a faster search, never had to face a unhealthy node and native tensor support (And Native ANN is coming soon https://github.com/vespa-engine/vespa/issues/9747 ). Very mature, and still progressing at a neck-break rate ( https://blog.vespa.ai )
Comment linking /
on “Elasticsearch from the Bottom Up (2013)”Submitted An open source scalable shopping back end w. navigation, search, recommendation as a story · 10 points
HN discussionFor use in retrieval/RAG, an emerging paradigm is to not parse the PDF at all. By using a multi-modal foundation model, you convert visual representations ("screenshots") of the pdf directly into searchable vector representations. Paper: Efficient Document Retrieval with Vision Language Models - https://arxiv.org/abs/2407.01449 Vespa.ai blog post https://blog.vespa.ai/retrieval-with-vision-language-models-... (my da…
Comment linking /retrieval-with-vision-language-models-colpali/
on “Ask HN: What are you using to parse PDFs for RAG?”Submitted Scalable realtime blog recommendation using neural nets with Vespa as a story · 9 points
HN discussionYes, https://blog.vespa.ai/vespa-hybrid-billion-scale-vector-sear...
Comment linking /vespa-hybrid-billion-scale-vector-search/
on “Solr’s Dense Vector Search for indexing and searching dense numerical vectors”Submitted Benchmark: Scaling TensorFlow model evaluation using Vespa as a story · 7 points
HN discussionI got great results with Vespa.ai https://blog.vespa.ai/using-approximate-nearest-neighbor-sea...
Comment linking /using-approximate-nearest-neighbor-search-in-real-world-applications/
on “Vald: A Highly Scalable Distributed Vector Search Engine”Submitted Introducing TensorFlow support in Vespa as a story · 7 points
HN discussionhttps://twitter.com/eugeneyan/status/1678060204943097863 >When Deepmind needs semantic retrieval, they just use the largest index on the planet. Fun fact: Query-doc similarity was done via simple TF-IDF instead of vectors. It performed better than vector retrieval when retrieve docs > 45 (they used 50). https://blog.vespa.ai/improving-zero-shot-ranking-with-vespa... >This case illustrates that in-domain effectivenes…
Comment linking /improving-zero-shot-ranking-with-vespa/
on “Gzip and KNN Outperforms Transformers on Text Classification”Submitted Vespa.ai is spinning out of Yahoo as a separate company as a story · 6 points
HN discussionHey everyone, I also work at Weaviate. Weaviate has implemented Hybrid Search because it helps with search performance in a few ways (Zero-Shot, Out-of-Domain, Continual Learning). Members of the Qdrant team are arguing against implementing Hybrid Search in Vector Databases with 3 main points that I believe are incorrect: 1. That there are not comparative benchmarks on Hybrid Search. 2. “Multi-Tool” systems are gene…
Comment linking /improving-zero-shot-ranking-with-vespa-part-two/
on “On Hybrid Search with Qdrant”Submitted Perplexity builds AI Search at scale on Vespa.ai as a story · 5 points
HN discussionDisclaimer, I work on Vespa. If you look for just pure vector similarity search, there are many alternatives. But Vespa's tensor support, multi-vector indexing and the ability to express models like colBERT ( 1) or cross-encoders makes it stand out if you need to move beyond pure vector search support. Plus, for RAG use cases, it's a full blown text search engine as well, allowing hybrid ranking combinations. Also w…
Comment linking /pretrained-transformer-language-models-for-search-part-3/
on “Vespa.ai is spinning out of Yahoo as a separate company”Submitted Elasticsearch vs. Vespa Performance Comparison as a story · 5 points
HN discussionYeah, I worked with the Flickr team on that project. Scaling to billions of photos, with partial update support of popularity for ranking. Back then, the properties had to stand up their own Vespa cluster(s), later on we created a managed service out of it. And, yes, the original plan for Vespa was to be a Vertical Search Platform, that is where the name Vespa comes from. More on the history in this blog post https:…
Comment linking /vespa-is-becoming-its-own-company/
on “Vespa.ai is spinning out of Yahoo as a separate company”Submitted Real time distributed joins: Possible? Some times as a story · 5 points
HN discussionThey have absolutely the best take on embedding search + normal faceted search. Here is a blog post about how it works: https://blog.vespa.ai/vespa-hybrid-billion-scale-vector-sear...
Comment linking /vespa-hybrid-billion-scale-vector-search/
on “Vespa.ai is spinning out of Yahoo as a separate company”Submitted Scalable evaluation of ONNX models with open source Vespa as a story · 4 points
HN discussionYes, Vespa has a generic Tensor framework that allows you to index multiple vectors for a single field, see https://blog.vespa.ai/semantic-search-with-multi-vector-inde... for details. field embeddings type tensor<float>(p {}, x[384]) to represent a multi-vector field { "0": [0.1....], "1": [0.2,..] }
Comment linking /semantic-search-with-multi-vector-indexing/
on “Choosing vector database: a side-by-side comparison”Submitted Perplexity builds AI Search at scale on Vespa (Elasticsearch alternative) as a story · 3 points
HN discussionThanks for sharing. How does this compare with DiskANN ( https://zilliz.com/blog/diskann-a-disk-based-anns-solution-w... ) or HNSW-IF ( https://blog.vespa.ai/vespa-hybrid-billion-scale-vector-sear... )?
Comment linking /vespa-hybrid-billion-scale-vector-search/
on “On-disk HNSW index for Postgres with pg_embedding”Submitted Elasticsearch vs. Vespa Performance Comparison as a story · 3 points
HN discussion
