TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Sqlite-vec: Work-in-progress vector search SQLite extension that runs anywhere

314 pointsby brylie10 months ago

15 comments

alexgarcia-xyz10 months ago
Author here, happy to answer any questions! Been working on this for a while, so I&#x27;m very happy to get this v0.1.0 &quot;stable&quot; release out.<p>sqlite-vec works on MacOS, Linux, Windows, Raspberry Pis, in the browser with WASM, and (theoretically) on mobile devices. I focused a lot on making it as portable as possible. It&#x27;s also pretty fast - benchmarks are hard to do accurately, but I&#x27;d comfortable saying that it&#x27;s a very very fast brute-force vector search solution.<p>One experimental feature I&#x27;m working on: You can directly query vectors that are in-memory as a contiguous block of memory (ie NumPy), without any copying or cloning. You can see the benchmarks for that feature here under &quot;sqlite-vec static&quot;, and it&#x27;s competitive with faiss&#x2F;usearch&#x2F;duckdb <a href="https:&#x2F;&#x2F;alexgarcia.xyz&#x2F;blog&#x2F;2024&#x2F;sqlite-vec-stable-release&#x2F;index.html#benchmarks" rel="nofollow">https:&#x2F;&#x2F;alexgarcia.xyz&#x2F;blog&#x2F;2024&#x2F;sqlite-vec-stable-release&#x2F;i...</a>
评论 #41144962 未加载
评论 #41143562 未加载
评论 #41143590 未加载
simonw10 months ago
Lots more details in Alex&#x27;s blog post here: <a href="https:&#x2F;&#x2F;alexgarcia.xyz&#x2F;blog&#x2F;2024&#x2F;sqlite-vec-stable-release&#x2F;index.html" rel="nofollow">https:&#x2F;&#x2F;alexgarcia.xyz&#x2F;blog&#x2F;2024&#x2F;sqlite-vec-stable-release&#x2F;i...</a>
评论 #41140003 未加载
Cieric10 months ago
I feel like I&#x27;ve touched a lot of things where something like this is useful (hobby projects). In my case I&#x27;ve done a recommendation engine, music matching (I specifically use it for matching anime to their data), and perceptual hash matching.
评论 #41140581 未加载
评论 #41141690 未加载
cotega9 months ago
I absolutely love this, great work! For those that might find it useful, I created a Python notebook that shows how to extend this to perform Hybrid Search (Vector + BM25 based Full Text search) <a href="https:&#x2F;&#x2F;github.com&#x2F;liamca&#x2F;sqlite-hybrid-search">https:&#x2F;&#x2F;github.com&#x2F;liamca&#x2F;sqlite-hybrid-search</a>
pjot10 months ago
I’ve done something similar, but using duckDB as the backend.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;patricktrainer&#x2F;duckdb-embedding-search">https:&#x2F;&#x2F;github.com&#x2F;patricktrainer&#x2F;duckdb-embedding-search</a>
评论 #41140484 未加载
bodantogat10 months ago
This sounds useful (I do a lot of throw-away text analysis on my laptop)
评论 #41140357 未加载
1yefuwang19 months ago
Hi, nice work. I write a similar vector search extension <a href="https:&#x2F;&#x2F;github.com&#x2F;1yefuwang1&#x2F;vectorlite">https:&#x2F;&#x2F;github.com&#x2F;1yefuwang1&#x2F;vectorlite</a> inspired by sqlite-vss using C++17 and hnswlib.<p>I&#x27;d like to do a benchmark to compare it with sqlite-vec, but I guess it is not a fair comparison given that sqlite-vec uses brute-force only.<p>One thing I&#x27;d recommend is to include recall rate in your benchmark data.<p>Brute force approach is a good starting point but doesn&#x27;t scale with serious production workload.
dang10 months ago
Related:<p><i>I’m writing a new vector search SQLite Extension</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=40243168">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=40243168</a> - May 2024 (85 comments)
deepsquirrelnet10 months ago
I love this. I know how much work addressing the dependencies must be, but you’re really attacking the right problems. Looking forward to trying this out with my project.
huevosabio10 months ago
Been using this for video games and it&#x27;s absolutely awesome. Alex, the author, is also great and very approachable.<p>I&#x27;ve been looking for something like this for a while.
评论 #41146137 未加载
nattaylor10 months ago
I have a use case for this that I&#x27;m excited to try. I&#x27;m glad AlexG has put so much effort into this. Even the docs are pretty good!<p>My pyenv python3.12.2&#x27;s sqlite won&#x27;t load extensions even after installing with what I think are the correct command line flags. Argh!<p>My brew installed python3.12&#x27;s sqlite will load extensions though, so I can proceed.
mic4710 months ago
Nice. Been waiting for this release to try it out.
pietz10 months ago
Is this also what turso uses in their &quot;AI feature&quot;?
评论 #41140447 未加载
haolez10 months ago
What&#x27;s the maximum vector size?
评论 #41141454 未加载
fsndz10 months ago
I love this. I am currently doing this tutorial of RAG where the vector DB is simply postgreSQL and pgvector. I guess I can try to reproduce that with SQLite and sqlite-vec now ! Awesome: <a href="https:&#x2F;&#x2F;www.lycee.ai&#x2F;courses&#x2F;91b8b189-729a-471a-8ae1-717033c77eb5&#x2F;chapters&#x2F;a8494d55-a5f2-4e99-a0d4-8a79549c82ad" rel="nofollow">https:&#x2F;&#x2F;www.lycee.ai&#x2F;courses&#x2F;91b8b189-729a-471a-8ae1-717033c...</a>
评论 #41144609 未加载