# RyanCodrai/turbovec

A vector index built on TurboQuant, written in Rust with Python bindings

Repository: https://github.com/RyanCodrai/turbovec
Canonical: https://ross.abutalabs.com/products/turbovec
Homepage: https://pypi.org/project/turbovec/
Language: Rust
License: MIT
License Family: permissive
Topics: ann, avx512, embeddings, faiss, nearest-neighbor, neon, python, quantization, rag, rust, simd, vector-search, embedding, quant, turboquant
Last push: 2026-08-21T11:42:17+00:00

## Health v2 (maintenance only)
Score: 59/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 35, longevity 11
- inputs: {"age_days": 160, "days_push": 12, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 16441, forks 1426 (observed 2026-08-28T04:11:16.395445+00:00)

## What it is
turbovec is a Rust vector index with Python bindings implementing Google's TurboQuant algorithm for 2-4 bit vector quantization with SIMD-accelerated nearest-neighbor search. It supports online ingest without a training phase, incremental crash-safe saves, and filtered search, claiming faster search than FAISS IndexPQFastScan.

## Use cases
- search embeddings from millions of documents in limited RAM
- build a fully local air-gapped RAG stack
- quantize vectors to 2-4 bits without a training step
- run fast nearest-neighbor search with SIMD on ARM or x86
- incrementally persist a growing vector index
- filter search results by id allowlist without recall loss

## When to choose
- you need memory-efficient vector search with heavy quantization
- you want a pure local, self-hosted alternative to FAISS or managed vector services
- your corpus grows continuously and you want online ingest without retraining
- you need fast filtered ANN search on AVX-512 or NEON hardware

## When to avoid
- you need full float32 precision or exact nearest-neighbor search
- you require a distributed or managed vector database
- you need rich index types like HNSW graph search
- your hardware lacks SIMD support and scalar fallback speed is insufficient

## Facets
- artifact type: library
- maturity: active
- function: vector-database, search-engine, machine-learning, rag
- domain: machine-learning, artificial-intelligence, databases
- platform: python, rust, cross-platform
- tags: ann, quantization, simd, embeddings, nearest-neighbor, faiss-alternative, avx512, neon, retrieval-augmented-generation, search

## Member repositories
- RyanCodrai/turbovec (main) score 59

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:16.395445+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:05:06.201028+00:00, confidence not recorded.
  - readme: https://github.com/RyanCodrai/turbovec (fetched 2026-08-28T04:11:16.395445+00:00, sha e4e8e215e646)
  - homepage: https://pypi.org/project/turbovec/ (fetched 2026-08-29T08:02:18.417757+00:00, sha 4b4e8fead74a)
  - registry_crates: https://crates.io/api/v1/crates/turbovec (fetched 2026-08-29T08:02:18.427303+00:00, sha 4210f3e48cb6)
- Data as of 2026-08-30T08:39:29.467469+00:00.