Chroma vs zvec
Open-source embedding database for building AI apps with retrieval. — versus — Alibaba's open-source in-process vector database: billion-scale similarity search embedded in your app, with DiskANN on-disk indexing, native full-text search and hybrid retrieval.
Same job — the embedding store under a RAG pipeline. Chroma is the Python-native developer default; Zvec is the in-process C++ engine betting on raw speed, DiskANN memory economics and built-in hybrid retrieval.
| Chroma | zvec | |
|---|---|---|
| Stars | 29k | 15k |
| Forks | 2.4k | 931 |
| Language | Python | C++ |
| License | Apache-2.0 | Apache-2.0 |
| Last activity | 2 days ago | yesterday |
| Topics | rag, memory | rag |
| Curated connections | 8 | 1 |
Chroma — the curator's take
Open-source embedding database for building AI apps with retrieval.
zvec — the curator's take
The 'SQLite of vector search' position, executed with Alibaba-scale engineering: runs inside your process (no server to operate), DiskANN keeps memory flat at large scale, and v0.5 added native FTS + hybrid retrieval so one MultiQuery spans dense, sparse, filters and text — no external search engine. Battle-tested internally before open-sourcing. NOT for multi-service architectures needing a shared, network-accessible store with auth and replication — in-process is the whole point and the whole limitation; C++ core with Python/Go/Rust SDKs, so debugging beneath the binding is not for everyone.