Skip to main content
Zvec is distributed as pre-built binary packages for Python and Node.js. Installation is straightforward and requires no compilation.

​
Prerequisites

​
Python Requirements

  • Python version: 3.10, 3.11, or 3.12
  • Dependencies: NumPy >= 1.23 (automatically installed)

​
Supported Platforms

Zvec provides pre-built binaries for the following platforms:
  • Linux: x86_64 (AMD64) and ARM64 (aarch64)
  • macOS: ARM64 (Apple Silicon)
Windows support is not currently available. For unsupported platforms, see Building from Source.

​
Installation

1

Choose your language

Zvec supports both Python and Node.js. Select your preferred language below.
2

Install the package

pip install zvec
The installation typically completes in seconds since pre-built binaries are used.
3

Verify installation

Confirm that Zvec is installed correctly:
import zvec
print(f"Zvec version: {zvec.__version__}")
If you see the version number (Python) or success message (Node.js), you’re ready to go!

​
Platform-Specific Notes

​
Linux

Zvec binaries are built for manylinux_2_28, which is compatible with:
  • Ubuntu 22.04+ / Debian 11+
  • RHEL 9+ / CentOS Stream 9+ / Rocky Linux 9+
  • Amazon Linux 2023+
  • Other distributions with glibc 2.28+
For older distributions (glibc < 2.28), you’ll need to build from source.

​
macOS

Zvec supports macOS 11.0 (Big Sur) and later on Apple Silicon (ARM64).
Intel-based Macs (x86_64) are not currently supported with pre-built binaries. Please build from source or use Rosetta 2 emulation (may have performance impact).

​
Troubleshooting

​
Python Version Mismatch

Problem: pip install zvec fails with “No matching distribution found” Solution: Verify your Python version is between 3.10 and 3.12: 
python --version
If you have multiple Python versions, use the specific version: 
python3.11 -m pip install zvec

​
Platform Not Supported

Problem: Installation fails with “unsupported platform” or similar error Solution: Check your platform: 
# Linux
uname -m  # Should output: x86_64 or aarch64

# macOS
uname -m  # Should output: arm64
If your platform is not supported, see Building from Source.

​
Import Error on Linux

Problem: ImportError: libstdc++.so.6: version 'GLIBCXX_X.X.XX' not found Solution: Your system’s C++ standard library is outdated. Options:
  1. Update your system (recommended): 
    # Ubuntu/Debian
sudo apt update && sudo apt upgrade

# RHEL/CentOS/Rocky
sudo dnf update
  2. Build from source on your specific system

​
NumPy Compatibility Issues

Problem: Version conflicts with NumPy Solution: Zvec requires NumPy >= 1.23. Update NumPy: 
pip install --upgrade numpy
If you have strict NumPy version requirements, ensure they’re compatible with >= 1.23.

​
Virtual Environments

We recommend using virtual environments to avoid dependency conflicts: 
# Create virtual environment
python -m venv zvec-env

# Activate (Linux/macOS)
source zvec-env/bin/activate

# Activate (Windows)
zvec-env\Scripts\activate

# Install Zvec
pip install zvec

​
Optional Dependencies

​
For Machine Learning Integration

If you plan to use Zvec with machine learning frameworks, you may want to install: 
# PyTorch (for model inference)
pip install torch

# Transformers (for Hugging Face models)
pip install transformers sentence-transformers

# OpenAI (for embeddings via API)
pip install openai
These dependencies are optional. Zvec works with any embedding source, including pre-computed vectors.

​
Building from Source

For platforms without pre-built binaries or if you need custom build options, see the Building from Source guide.

​
Next Steps

Quickstart

Build your first vector search application

Core Concepts

Learn about schemas, collections, and indexing

Build docs developers (and LLMs) love

Get started for freeTalk to us