@AYi_AInotes: Wow, Alibaba has directly open-sourced the vector database it has been using internally for years. The capability that Pinecone charges $70/month for, you can get for free with a single pip command. Billion-level vector recall in milliseconds without needing a separate service. From now on, those doing RAG and AI search no longer need to pay Pinecone $70 each month! The vector database that Alibaba has been running internally for years is open-sourced...
Summary
Alibaba has open-sourced Zvec, a vector database used internally for years. It supports billion-scale vector retrieval in milliseconds, can be embedded into application processes without a separate service, and is completely free. It serves as a replacement for paid services like Pinecone.
View Cached Full Text
Cached at: 06/21/26, 04:33 AM
Wow, Alibaba just open-sourced their internally-used vector database, and it delivers the same capability as Pinecone’s $70/month plan with just a single pip install for free — billion-scale vector search in milliseconds, no separate service needed.
From now on, anyone building RAG or AI search can skip paying Pinecone $70 every month.
Alibaba’s long-running internal vector database is called Zvec, and you can start using it with one pip install — completely free.
Three of the most hardcore features:
- Billion-scale vector retrieval in milliseconds, no separate service required — embed it directly into your application process.
- Runs everywhere: servers, desktops, even Raspberry Pi.
- Official SDKs for all major languages. v0.5.0 adds native hybrid full-text search — vector + keyword filters in one query.
I think Alibaba just took their own production-grade wheel and gave it away to the entire industry. Now the infrastructure layer for AI applications has another free, reliable option.
pip install zvec
AYi (@AYi_AInotes): Humanity still can’t write the physical equation for a fried egg. Drop an egg into a hot oil pan — how it coagulates, spreads, and browns at the edge — no formula can describe it. This kind of example is countless in the physical world.
And that’s exactly the ceiling of the current general AI paradigm: video generation and VLA models all learn statistical correlations at the pixel level.
Similar Articles
@aronhouyu: Alibaba open-sourced a vector database zvec – no Docker needed, just pip install zvec. This is fun, although my database knowledge is still stuck at SELECT * FROM usertable. I'll try it out tonight.
Alibaba has open-sourced the vector database zvec, which can be installed directly via pip – lightweight and easy to use.
@HowToPrompt__: China open-sourced a vector database that destroys Pinecone, Chroma, and Weaviate. It's called Zvec, an in-process vect…
China open-sourced Zvec, an in-process vector database that runs inside apps without servers, supporting billions of vector searches in milliseconds and battle-tested at Alibaba scale.
@shedntcare_: BREAKING: Alibaba just dropped a vector database that could change RAG forever. Meet Zvec No server. No Docker. No clou…
Alibaba released Zvec, a fully open-source vector database (Apache 2.0) that can be installed via pip and supports dense, sparse, and hybrid search for RAG applications, processing billions of vectors in milliseconds.
@vintcessun: Compressing 10 million vectors from 31GB to 4GB, with search even faster than FAISS — sounds crazy, but Turbovec actually did it. The core is Google's TurboQuant data-independent quantization: no training, no parameter tuning, just add vectors and index. Handwritten NEON/AVX-512 implementations are genuinely 12-20% faster, supporting filtered search by ID, saving a ton of post-processing hassle. Rust under the hood + pip install, minimal maintenance cost.
Turbovec, based on Google's TurboQuant algorithm, compresses 10 million vectors from 31GB to 4GB, with search speed 12-20% faster than FAISS, supports filtered search, and offers a Rust implementation with a Python package.
@AriXZone: The Linux kernel, a very large project with 28 million lines of code and 75,000 files, takes only about 3 minutes to index, with sub-millisecond query responses. codebase-memory-mcp is an open-source MCP server from DeusData that provides AI coding assistants (Claude Cod…
DeusData has open-sourced an MCP server called codebase-memory-mcp, which pre-parses code repositories into persistent knowledge graphs, providing AI coding assistants with sub-millisecond code structure query capabilities. It supports 158 languages, reduces token consumption by approximately 99% compared to traditional grep, and has extremely fast indexing speed.