conda vs uv
conda
conda.io
uv
docs.astral.shSide by side
| Figure | conda | uv |
|---|---|---|
| Price from | — | Free |
| Free plan | — | Yes |
| Free trial | — | — |
| Open source | — | — |
| Windows | Yes | Yes |
| macOS | Yes | Yes |
| Linux | Yes | Yes |
| Plans published | 0 | 1 |
| Spec score · Python Package Managers | 7.8 (#12) | 8.0 (#10) |
| Spec score · Package Managers | 7.8 (#43) | 8.9 (#4) |
| Dependency resolution | Yes | Yes |
| Free plan | Yes | Yes |
| Integrations | Conda can work with pip in an active conda environment, though the documentation notes unavoidable compatibility limits between pip and conda packages. | The documentation lists integrations and guides for Docker, Jupyter, marimo, GitHub Actions, GitLab CI/CD, Pre-commit, PyTorch, FastAPI, and several package registries. |
| Lockfile support | Yes | Yes |
| Offline installation | Yes | Yes |
| Package formats | Conda packages can contain system libraries, language modules, executable programs, and metadata, and use the .conda or .tar.bz2 file formats. | source distributions,wheels,requirements.txt,pylock.toml |
| Private registry auth | Yes | Yes |
| Purpose | Conda provides package, dependency, and environment management for any language. | uv is an extremely fast Python package and project manager written in Rust. |
| Supported platforms | Windows,macOS,Linux | macOS,Linux,Windows,CPython,PyPy,Pyodide,GraalPy |
Plans
In their own words
uv
uv is a Python package and project manager for Linux, macOS, and Windows. Written in Rust, it manages dependencies and environments, with lockfiles and workspaces, and can install, list, pin, or uninstall Python versions. It also runs scripts with dependencies and runs or installs command-line tools from Python packages. Its pip-compatible interface is intended to cover work done by tools such as pip, pip-tools, pipx, poetry, pyenv, twine, and virtualenv, though its commands do not exactly match those tools. The documentation describes uv as 10–100x faster than pip. It supports source distributions, wheels, requirements.txt, and pylock.toml, as well as offline installation and private registry authentication. For package selection, its default first-index strategy limits candidates to the first index where a package is found. Credentials are not stored in uv.lock, but uv currently stores them in a plaintext file; native system secret storage is an experimental preview. The free plan is free.