Quantum Programming Frameworks, the 2026 Guide

Choosing where to start is half the battle in quantum software. There are now many quantum programming frameworks to pick from, and the landscape has shifted enough since we first wrote this guide that the old shortlist no longer tells the whole story. This is our 2026 update, covering the frameworks that matter today and how to work out which one fits your project.

The tools below span every major approach, from IBM’s gate-model Qiskit to D-Wave’s annealing toolkit and NVIDIA’s GPU-accelerated platform. Most are open source, most are written in Python, and all of them run on classical simulators, so you can learn the basics long before you touch real hardware.

Quantum Languages Versus Quantum Frameworks

It helps to separate two ideas that are often used interchangeably. A quantum programming language defines the syntax for describing quantum states, gates and measurements, much as a classical language does for ordinary code. Microsoft’s Q# is a clear example, a language built specifically for the job.

A quantum programming framework is the wider toolkit around that. It bundles circuit builders, simulators, optimisation routines, noise models and the connections to real hardware or cloud services. Most of the options here are frameworks in that broader sense, and several let you write ordinary Python while they handle the quantum machinery underneath.

The Main Quantum Programming Frameworks in 2026

We have grouped the eight frameworks below by the job they do best rather than ranking them, since a beginner’s first circuit and a supercomputing pipeline pull in very different directions. For a sense of how usage has shifted over time, our earlier look at the popularity of quantum languages and frameworks gives the longer view.

Qiskit

Qiskit, IBM’s open-source Python SDK, is the most widely used framework for gate-model quantum computing. It reached version 1.0 in February 2024, which trimmed the package down to a standalone core, and the fast-moving 2.x line has carried on through 2026, reaching version 2.5 by mid-year. Older tools such as Aqua had already been broken out into separate application libraries a few years earlier. It runs on IBM’s cloud machines and simulators, and its large community makes it the usual first stop for newcomers.

Cirq

Cirq, from Google Quantum AI, is another Python framework built around near-term noisy circuits. It ships new versions on a roughly quarterly cadence and gives fine control over qubit layout and timing, which suits researchers targeting a specific processor. Google’s own hardware and virtual-machine simulators are first-class targets.

PennyLane

PennyLane, built by the Canadian company Xanadu, is the leading framework for quantum machine learning and differentiable quantum programming. It treats a quantum circuit as a function you can take gradients through, so it slots neatly into machine-learning workflows, and it is hardware agnostic, running across most other vendors’ backends. Its companion Catalyst compiler adds just-in-time compilation for heavier programs.

Microsoft Q# and the Azure Quantum Development Kit

Microsoft takes a different route with Q#, a dedicated quantum language rather than a Python library, paired with the Azure Quantum Development Kit. Microsoft rewrote that kit from scratch in early 2024, and the modern version is faster, runs in the browser and drops the old Visual Studio and C# dependencies. The classic kit is now deprecated, so older Q# projects need porting to the new one.

Amazon Braket SDK

The Amazon Braket SDK is the Python interface to Amazon Web Services’ managed quantum service, and its main appeal is breadth. From one codebase it reaches superconducting, trapped-ion and neutral-atom machines from several hardware providers, alongside a range of simulators. That makes it a practical choice for teams that want to compare devices without signing up to each vendor separately.

NVIDIA CUDA-Q

CUDA-Q is NVIDIA’s platform for hybrid quantum-classical computing, renamed from CUDA Quantum in April 2024. It is built for performance, using graphics cards to simulate large circuits and to orchestrate workloads that span processors, GPUs and real quantum devices. For anyone treating quantum computing as part of a high-performance computing pipeline, it is the most prominent option to have appeared since 2023.

pytket and TKET

TKET, used through the Python package pytket, is Quantinuum’s circuit compiler, and its purpose is optimisation rather than owning hardware. It rewrites and streamlines circuits so they run better on whatever backend you target, and it connects to many of them. Recent versions handed job submission to a separate package called qnexus, leaving pytket focused on compilation and local emulation.

D-Wave Ocean

Ocean is D-Wave’s Python toolkit, and it stands apart because it targets quantum annealing rather than the gate model the others use. Annealing suits optimisation problems, and Ocean also offers hybrid solvers that combine classical and quantum resources for larger cases. It is the toolkit to reach for when a problem looks like a scheduling or constraint puzzle rather than a general algorithm.

Quantum Programming Frameworks Compared

Other Frameworks Worth Knowing

Two more names come up often. Rigetti’s pyQuil, built on the Quil instruction language, is still actively maintained and is the way into Rigetti’s superconducting hardware, though it is tied to that one vendor. TensorFlow Quantum, which blends Cirq with TensorFlow, went quiet for a stretch and then resumed regular releases, and it remains a niche choice for researchers combining machine learning with quantum circuits.

Frameworks That Have Wound Down

A couple of once-notable frameworks are no longer live choices. Strawberry Fields, Xanadu’s photonic and continuous-variable toolkit, was archived in January 2026 and is now read-only, accepting only bug fixes. Xanadu has since centred its software effort on PennyLane, which carries its own continuous-variable support, so new photonic projects are better started there. ProjectQ, one of the earliest open frameworks when it arrived around 2016, has had no release since 2022 and is best treated as history rather than a starting point.

How to Choose a Quantum Framework

There is no single best framework, only the best fit for a task. Beginners usually start with Qiskit or PennyLane because both are Python-first and thoroughly documented, and both run on free simulators. If you would rather learn a purpose-built language, Q# is the one to try.

The choice narrows quickly once you know the problem. Teams working on optimisation lean towards D-Wave’s Ocean, those building models favour PennyLane, and anyone folding quantum work into high-performance computing looks at CUDA-Q. When the priority is comparing real hardware, Amazon Braket reaches the widest set of machines from one place, and if you prefer to go old school, our roundup of quantum textbooks covers the theory behind all of it.

Frequently Asked Questions

What is the most popular quantum programming framework? By community size and adoption it is Qiskit, now on its 2.x line, with Cirq and PennyLane the next most widely used. Popularity shifts year to year, but the Python-based frameworks consistently lead.

Which quantum programming framework should a beginner learn first? Qiskit and PennyLane are the easiest places to start, since both use Python, have strong documentation and run on free simulators. You can build and test real circuits on a laptop before ever booking time on quantum hardware.

Do quantum programming frameworks need real quantum hardware? No. Every framework here ships with classical simulators that run on an ordinary computer, which is how most learning and development happens. You only need cloud access to a real device when you want to run a finished program on quantum hardware.