Anaconda now supports Windows on Arm natively. If you’re on a Snapdragon laptop, a Surface device or another Windows Arm64 PC, you can install conda and run Python packages built for your processor. Before this, they ran through x64 emulation.
What’s available
Miniconda 26.9.3 for Windows Arm64
Native installers with Python 3.12, 3.13 and 3.14.
Download Miniconda · Release notes
Native packages in our main channel, including:
- Data science: NumPy, SciPy, pandas, XGBoost, Numba
- AI and GPU: PyTorch 2.14 (CPU and CUDA builds), the CUDA Toolkit, llama.cpp (CPU and CUDA builds)
- Web and data access: Flask, Django, cryptography, psycopg, pyodbc
See every available package in the win-arm64 channel index.
Kilo Desktop for Windows Arm
Kilo Desktop runs natively on Windows Arm. It’s an agentic AI coding assistant for software and data science. You get access to 500+ models, agents that plan and write code, and notebooks the agent can edit directly. On Windows Arm, it also manages conda environments from Anaconda’s packages and runs open models locally on your device.
Getting started
Download the Windows Arm64 Miniconda installer from anaconda.com/download, then see the docs:
- Windows Arm guide: choosing an installer, running x64 environments alongside native ones, migrating existing environments and troubleshooting
- Installing Miniconda on Windows: step-by-step install
- Miniconda release notes: what’s in 26.9.3
- win-arm64 channel index: the full package list
Built with partners
This launch was a joint effort with Microsoft and Qualcomm. Read more:
- Microsoft: Python 3.15 delivers a native-by-default experience on Windows on Arm
- Qualcomm: Native Python on Windows on Snapdragon
- Anaconda: Python 3.15: What Data Scientists and Developers Need to Know
- Anaconda: Anaconda on Microsoft Surface Laptop Ultra
Tell us what you need
Reply here with:
- Packages you need that aren’t on Windows Arm64 yet
- Issues you run into
- What you’re building
Your feedback decides what we build next.