Skip to main contentIBM Quantum Documentation

Install Qiskit

Whether you will work locally or in a cloud environment, the first step for all users is to install Qiskit. For those wanting to run on a real system, your next step is to choose one of two channels in order to access IBM Quantum™ systems: IBM Quantum Platform or IBM Cloud®.

Upgrade from Qiskit 0.x to Qiskit 1.0 and beyond

(If you are installing Qiskit for the first time, skip ahead to the Install and set up section. This notice is relevant only to users who have installed Qiskit previously.)

For those upgrading from version 0.x to 1.0 or later: note that because Qiskit 1.0 uses a new packaging structure, you cannot use pip install -U qiskit to upgrade from any Qiskit 0.x version to 1.0.

See the Qiskit 1.0 migration guide for details and instructions.

Future updates starting with Qiskit 1.0 will allow for in-place upgrades.

Install the Qiskit SDK and the Qiskit Runtime client

  1. Install Python. Check the "Programming Language" section on the Qiskit PyPI project page(opens in a new tab) to determine which Python versions are supported by the most recent release. For download instructions, see the Python Beginners Guide.(opens in a new tab)

    We recommend that you use Python virtual environments(opens in a new tab) to separate Qiskit from other applications.

    If you're new to virtual environments, click here for more information.

    A virtual Python environment is an isolated space to work with Python for a specific purpose — so you can install whatever packages you wish, and set up libraries, dependencies, and so on, without affecting the "base" Python environment on your machine.

    One important advantage of a virtual environment is that if your Python environment becomes corrupted somewhere along the way, you can easily delete the virtual environment and start over!

    Choose a preferred location in which to store information about your virtual environments. Commonly they're stored in a directory named .venv within a user's home directory.

    First, create a minimal environment with only Python installed in it.

    python3 -m venv /path/to/virtual/environment

    Next, activate your new environment.

    source /path/to/virtual/environment/bin/activate
  2. Install pip(opens in a new tab). Pip is a Python package manager that you use to install Qiskit and other Python packages. Use pip list to see what is in your virtual environment.

  3. Install the Qiskit SDK. If you plan to run jobs on quantum hardware, also install Qiskit Runtime.

    pip install qiskit
    pip install qiskit-ibm-runtime

    If you intend to use visualization functionality or Jupyter notebooks, it is recommended to install Qiskit with the extra visualization support ('qiskit[visualization]').

    pip install qiskit[visualization]
  4. If you want to run a Jupyter notebook with the Qiskit packages you just installed, you will need to install Jupyter in your environment.

    pip install jupyter

    Then open your notebook as follows:

    jupyter notebook path/to/notebook.ipynb

If you are planning to work locally and use simulators built into Qiskit, then your installation is done. If you want to run jobs on IBM Quantum systems, next select an access channel and finish your setup.

Stay current with the latest versions

Periodically check the Qiskit release notes and the Qiskit Runtime release notes to see new releases. We recommend frequently updating your requirements for qiskit and qiskit-ibm-runtime by, for example, changing the versions in requirements.txt to the latest versions.


"No Module 'qiskit'" error with Jupyter Notebook

If you used pip install qiskit and set up your virtual environment in Anaconda, then you may get the No Module 'qiskit' error when you run a tutorial in Jupyter Notebook. If you have not installed Qiskit or set up your virtual environment, you can follow the installation steps.

The error is caused when trying to import the Qiskit package in an environment where Qiskit is not installed. If you launched Jupyter Notebook from the Anaconda-Navigator, it is possible that Jupyter Notebook is running in the base (root) environment, instead of in your virtual environment. Choose a virtual environment in the Anaconda-Navigator from the Applications on dropdown menu. In this menu, you can see all of the virtual environments within Anaconda, and you can select the environment where you have Qiskit installed to launch Jupyter Notebook.

Compilation errors during installation

Qiskit depends on a number of other open-source Python packages, which are automatically installed when doing pip install qiskit. Depending on your system's platform and Python version, it is possible that a particular package does not provide pre-built binaries for your system. You can refer to Operating system support for a list of platforms supported by Qiskit, some of which may need an extra compiler. In cases where there are no precompiled binaries available, pip will attempt to compile the package from source, which in turn might require some extra dependencies that need to be installed manually.

If the output of pip install qiskit contains similar lines to:

Failed building wheel for SOME_PACKAGE
build/temp.linux-x86_64-3.5/_openssl.c:498:30: fatal error
compilation terminated.
error: command 'x86_64-linux-gnu-gcc' failed with exit status 1

please check the documentation of the package that failed to install (in the example code, SOME_PACKAGE) for information on how to install the libraries needed for compiling from source.

Operating system support

Qiskit strives to support as many operating systems as possible, but due to limitations in available testing resources and operating system availability, not all operating systems can be supported. Operating system support for Qiskit is broken into three tiers with different levels of support for each tier. For platforms outside these, such as FreeBSD or WebAssembly (WASI), Qiskit may still be installable, but it is not tested and you will have to build Qiskit (and likely Qiskit’s dependencies) from source.

Additionally, Qiskit only supports the CPython implementation of the Python language. Running with other Python interpreters such as PyPy is not supported.

Tier 1

Tier 1 operating systems are fully tested as part of the development processes to ensure any proposed change will function correctly. Pre-compiled binaries are built, tested, and published to PyPI as part of the release process. Typically, as long as there is a functioning Python environment installed, Qiskit can be installed on these operating systems without needing to install further dependencies.

Tier 1 operating systems:

  • Linux x86_64 (distributions compatible with the manylinux 2014(opens in a new tab) packaging specification).
  • macOS x86_64 (10.12 or later)
  • macOS ARM64 (11.0 or newer)
  • Windows 64-bit (Windows 10 and later supported)

Tier 2

Tier 2 operating systems are not tested as part of development process. However, pre-compiled binaries are built, tested, and published to PyPI as part of the release process and these packages can be expected to be installed with just a functioning Python environment.

Tier 2 operating systems:

Tier 3

Tier 3 operating systems are not tested as part of the development process. Pre-compiled binaries are built and published to PyPI as part of the release process but are not tested. They may not be installable with just a functioning Python environment and might require a C/C++ compiler or additional programs to build dependencies from source as part of the installation process. Support for these operating systems are best effort only.

Tier 3 operating systems:

Qiskit versioning

Qiskit version numbers follow Semantic Versioning(opens in a new tab). The version number is comprised of three primary components: the major, minor, and patch versions. For example, in version number X.Y.Z, X is the major version, Y is the minor version, and Z is the patch version.

Breaking API changes are reserved for major version releases. The minimum period between major version releases is one year. Minor versions introduce new features and bug fixes without breaking API compatibility, and are periodically (currently every three months) published for only the current major version. Patch versions provide fixes for bugs identified in the most recent minor version of each actively supported release series (that is, the major version). We support at most two release series at a time, which occurs only during the period of overlap following a new major version release, described in more detail below.

Release schedule

A tentative release schedule is included below:

Tentative Qiskit release schedule

For an up-to-date release schedule, refer to the Qiskit Github project's milestones list(opens in a new tab), which will always contain the current release plan.

With the release of a new major version, the previous major version is supported for at least six months; only bug and security fixes are accepted during this time and only patch releases are published for this major version. A final patch version is published when support is dropped, and that release also documents the end of support for that major version series. A longer support window is needed for the previous major version as this gives downstream Qiskit consumers and their users a chance to migrate their code. Downstream libraries that depend on Qiskit should not raise their minimum required Qiskit version to a new major version immediately after its release because the library's user base needs time to migrate to the new API changes. Having an extended support window for the previous major Qiskit version gives downstream projects time to ensure compatibility with the next major version. Downstream projects can provide support for two release series at a time to give their users a migration path.

For the purposes of semantic versioning, the Qiskit public API is considered any documented module, class, function, or method that is not marked as private (with an underscore _ prefix). However, there can be explicit exceptions made for specific documented APIs. In such cases, these APIs will be clearly documented as not being considered stable interfaces yet, and a user-visible warning will be actively emitted on any use of these unstable interfaces. Additionally, in some situations, an interface marked as private is considered part of the public API. Typically this only occurs in two cases: either an abstract interface definition where subclasses are intended to override/implement a private method as part of defining an implementation of the interface, or advanced-usage low-level methods that have stable interfaces but are not considered safe to use, as the burden is on the user to uphold the class/safety invariants themselves (the canonical example of this is the QuantumCircuit._append method).

The supported Python versions, minimum supported Rust version (for building Qiskit from source), and any Python package dependencies (including the minimum supported versions of dependencies) used by Qiskit are not part of the backwards compatibility guarantees and may change during any release. Only minor or major version releases will raise minimum requirements for using or building Qiskit (including adding new dependencies), but patch fixes might include support for new versions of Python or other dependencies. Usually the minimum version of a dependency is only increased when older dependency versions go out of support or when it is not possible to maintain compatibility with the latest release of the dependency and the older version.

Upgrade strategy

When a new major version is released, the recommended upgrade path is to first upgrade to the most recent minor version on the previous major version. Shortly before a new major version, a final minor version will be published. This final minor version release X.Y+1.0.0 is equivalent to X.Y.0 but with warnings and deprecations for any API changes that are made on the new major version series.

For example, immediately proceeding the 1.0.0 release, a 0.46.0 release will be published. The 0.46.0 release will be equivalent to the 0.45.0 release but with additional deprecation warnings that document the API changes that were made as part of the 1.0.0 release. This pattern will be used for any future major version releases.

Qiskit users should first upgrade to this final minor version to see any deprecation warnings and adjust their Qiskit usage before trying a potentially breaking release. The previous major version will be supported for at least six months to give sufficient time to upgrade. A typical pattern to manage this is to pin the maximum version to avoid using the next major release series until you're sure of compatibility. For example, specifying qiskit<2 in a requirements file when the current major Qiskit version is 1 ensures that you're using a version of Qiskit that doesn't have breaking API changes.

Capping the version less than the next major version ensures that you see any deprecation warnings before a major version release. Without the cap, pip installs the newest version available by default.


For each minor and major version release, Qiskit publishes pre-releases that are compatible with PEP440(opens in a new tab). Typically these are release candidates of the form X.Y.0rc1. The rc releases will have a finalized API surface and are used to test a prospective release.

Note that when one of the PEP440 pre-release suffixes (such as a, b, or pre) are published, it does not have the same guarantees as an rc release, and is only a preview release. The API might change between these pre-releases and the final release with that version number. For example, 1.0.0pre1 might have a different final API than 1.0.0.


If there are issues with a release's packaging, a post-release might be issued to correct this. These will follow the form X.Y.Z.1 where the fourth integer indicates that it is the first post-release of the X.Y.Z release. For example, the qiskit-terra (the legacy package name for Qiskit) 0.25.2 release had some issue with the sdist package publishing, and a post-release was published that corrected this issue. The code was identical, and only fixed the packaging issue for the release.

Next steps

Was this page helpful?
Report a bug or request content on GitHub.