mirror of
https://github.com/hanzoai/vector.git
synced 2026-10-30 18:18:56 +00:00
zeekay
6e02ab9e1a
hanzoai/vector is a Qdrant fork — 375 files on `master` name qdrant, the history starts at Qdrant's own `init` commit, and Cargo.toml declares Apache-2.0. The code was fully rebranded to `hanzo-vector` with no NOTICE, so nothing in the distributed artifact named the upstream author. The NOTICE text is not new. z wrote it on 2026-06-21 in10dd93dfe("chore: restore upstream attribution + NOTICE (OSS compliance)"), but that commit landed on ci/canonical-docker-build-1776997064 and was never merged, so `master` — the branch that carries the code — never got it. This carries his wording over verbatim. What is deliberately NOT carried over:10dd93dfealso prepended two copyright lines to LICENSE. Upstream Qdrant's LICENSE is the bare Apache-2.0 text with no copyright line in it (the appendix still reads "Copyright [yyyy] [name of copyright owner]"), so there is no upstream notice there to retain — those lines would be an edit to an upstream licence, not a preservation of one. Attribution belongs in NOTICE, which is also what Apache-2.0 s4(d) asks for. LICENSE is untouched. Note for whoever owns the default branch: `main` is a four-file stub (README, LLM.md, CLAUDE.md, LICENSE) whose LICENSE reads "MIT License, Copyright (c) 2026 Hanzo AI". No Apache-licensed code sits under it, so it is not a redistribution violation today, but the default branch of a Qdrant fork asserting MIT is misleading and should be reconciled with master.
Hanzo Vector
High-performance vector search engine for the Hanzo AI platform. Based on Qdrant.
Quick Start
docker run -p 6333:6333 -p 6334:6334 ghcr.io/hanzoai/vector:latest
- REST API on port
6333 - gRPC API on port
6334
Features
- Dense and sparse vector search with filtering and payloads
- Hybrid search combining semantic and keyword matching
- Vector quantization (scalar, product, binary) for memory efficiency
- Distributed deployment with sharding and replication
- HNSW indexing with SIMD acceleration
- Write-ahead logging and snapshot-based persistence
- GPU acceleration (NVIDIA and AMD)
Configuration
Mount a custom config file:
docker run -p 6333:6333 \
-v $(pwd)/config.yaml:/qdrant/config/production.yaml \
-v $(pwd)/data:/qdrant/storage \
ghcr.io/hanzoai/vector:latest
See config/config.yaml for all available options.
Building from Source
cargo build --release --bin qdrant
API
- REST: OpenAPI 3.0 schema in
openapi/ - gRPC: Protocol definitions in
lib/api/src/grpc/proto/
Links
- Hanzo AI
- GitHub
- Qdrant Documentation (upstream docs)
License
Apache License, Version 2.0. See LICENSE.
Based on Qdrant -- vector similarity search engine.