The ecosystem around the SDKs — compilers, decoders, benchmark suites, visualizers and simulators built for one job each. 27 of the 29 tools here are open source, and every one of them is free to use.
Looking for the SDK layer instead? Start with the quantum SDKs, or compare them side by side. For general-purpose simulation, the free quantum simulators page is the better starting point — the simulators listed below are specialist tools.
An SDK gets you as far as a circuit that runs. What it rarely gets you is a circuit that runs well. The moment you touch real hardware, three separate problems appear: the circuit is deeper than the device can hold, the results are biased by noise, and you have no neutral way to tell whether the machine or your code is at fault. Each of those has its own category of tool below — transpilation, error mitigation, and benchmarking.
Further out sits quantum error correction, which is a different discipline with a different toolchain: you simulate stabilizer circuits, decode syndromes, and count what a logical qubit would actually cost. None of that lives in a mainstream SDK.
Last verified: 2026-07-19 — projects change; check the repo before depending on one.
Circuit builders & visualizers
Browser-based drag-and-drop circuit editor that simulates up to 16 qubits and animates amplitudes, Bloch vectors and measurement probabilities as you edit. Nothing to install — the whole tool is one page.
Use it when
Reach for it when you want to see what a gate actually does to a state before you write any code.
Background: quantum circuits
Circuit builders & visualizers
IBM's graphical circuit builder: drag gates onto a staff, watch the q-sphere and probability histogram update, then run the circuit on real IBM hardware and export it as OpenQASM or Python.
Use it when
Use it when you want a circuit you drew by hand to run on an actual QPU without setting up a local environment.
Circuit builders & visualizers
Interactive single-qubit Bloch sphere simulator in the browser. Apply Pauli, Hadamard and phase gates or arbitrary axis rotations and pulse sequences, then export a publication-quality figure.
Use it when
Reach for it when you are teaching or debugging single-qubit rotations and need a picture rather than a state vector.
Background: the Bloch sphere
Circuit builders & visualizers
The plotting layer inside the Qiskit SDK. `circuit_drawer()` renders circuits as text, Matplotlib or LaTeX, and state helpers such as `plot_bloch_multivector` and `plot_state_city` visualize the resulting quantum state.
Use it when
Use it whenever a circuit is misbehaving and you want to see what the transpiler actually produced, not what you thought you wrote.
Background: quantum circuits
Compilers & transpilers
Quantinuum's retargetable quantum compiler — a C++ core with `pytket` Python bindings — that optimizes gate counts, routes circuits onto a device's qubit connectivity and emits code for many QPUs and simulators.
Use it when
Reach for it when your SDK's built-in transpiler is leaving too many two-qubit gates in the circuit and depth is what is killing your results.
Background: transpilation
Compilers & transpilers
Berkeley Lab's compiler and synthesis framework. Rather than only rewriting gates, it partitions a circuit into blocks and numerically re-synthesizes each one from its unitary, often producing far shorter circuits than rule-based passes.
Use it when
Use it when you have a small-to-medium circuit that matters enough to spend real compute time shrinking.
Background: transpilation
Compilers & transpilers
A header-only C++ library and set of command-line tools that parse, optimize and device-map OpenQASM 2.0 circuits by manipulating the syntax tree directly. Fast, dependency-light and scriptable.
Use it when
Reach for it when you want circuit optimization inside a build pipeline or a C++ program rather than a Python notebook.
Error mitigation & QEC
The Unitary Foundation's framework-agnostic error-mitigation toolkit, implementing zero-noise extrapolation, probabilistic error cancellation, dynamical decoupling and more. It wraps circuits from Qiskit, Cirq, Braket, PennyLane, Qibo and raw OpenQASM 3.
Use it when
Use it when results from real hardware are systematically off and you want a principled correction instead of averaging harder.
Background: a practical guide to error mitigation
Error mitigation & QEC
Google Quantum AI's stabilizer-circuit simulator, fast enough to sample error-correction experiments with thousands of qubits and millions of operations. It also emits the detector error model that decoders consume.
Use it when
Reach for it the moment you start simulating a surface code — general state-vector simulators cannot reach these sizes.
Background: why decoding is the fault-tolerance bottleneck
Error mitigation & QEC
A minimum-weight perfect matching decoder ('sparse blossom') for surface codes and other QEC codes, written in C++ with a Python API and designed to take Stim's detector error models directly.
Use it when
Use it when you have syndrome data from Stim or hardware and need to turn it into an actual correction, quickly.
Background: why decoding is the fault-tolerance bottleneck
Error mitigation & QEC
A Qiskit-community framework for error-correction research — code construction, noisy circuit generation and decoders such as union-find. Listed for completeness, but the repository has seen no meaningful commits since early 2025.
Use it when
Only worth reading as reference code; for working QEC tooling today, Stim plus PyMatching is the maintained path.
Background: logical qubits and fault tolerance explained
High-performance simulators
Google's optimized state-vector simulator, using gate fusion, AVX/FMA vectorization, multithreading and optional GPU offload. It plugs into Cirq through the `qsimcirq` package.
Use it when
Reach for it when Cirq's built-in simulator has become the bottleneck and you need another handful of qubits.
Background: the best free quantum simulators
High-performance simulators
GPU-accelerated simulation libraries — cuStateVec, cuTensorNet and cuDensityMat — that other frameworks call under the hood. Qiskit Aer, qsim, PennyLane and CUDA-Q all have cuQuantum backends. The core libraries are free but proprietary; only the Python bindings and samples are open source.
Use it when
Use it when you have GPUs available and the simulation, not the algorithm, is what you are waiting on.
Background: quantum simulators compared
High-performance simulators
A C/C++ simulator of state vectors and density matrices that is multithreaded, MPI-distributed and GPU-accelerated from the same source, so the same program scales from a laptop to a supercomputer.
Use it when
Reach for it when your simulation needs to run across cluster nodes rather than on one machine.
Background: quantum simulators compared
High-performance simulators
Qiskit's C++ simulator backend, offering state-vector, density-matrix, stabilizer and matrix-product-state methods plus realistic device noise models. Its README now states the project is under reduced maintenance, taking critical fixes only.
Use it when
Use it to reproduce a specific device's noise before you spend real QPU time — but note the reduced-maintenance status when planning long-lived work.
Background: the best free quantum simulators
Benchmarking & characterization
A community platform from the Unitary Foundation that collects and publishes quantum device performance results, with `metriq-gym` as the runnable benchmark suite behind it. The platform is explicitly still in beta.
Use it when
Reach for it when a vendor claim needs a second opinion from results somebody else can reproduce.
Background: why quantum benchmarking became policy
Benchmarking & characterization
Infleqtion's scalable, application-oriented benchmark suite — GHZ, Mermin-Bell, QAOA, VQE, phase estimation and more — measuring devices on realistic workloads rather than isolated gate fidelities. It ships inside the `client-superstaq` monorepo.
Use it when
Use it when you need device numbers that reflect the kind of circuit you actually intend to run.
Background: why quantum benchmarking became policy
Benchmarking & characterization
The Quantum Economic Development Consortium's benchmark programs — Bernstein-Vazirani, QFT, Grover, VQE, QAOA, HHL, Hamiltonian simulation — measuring result quality and execution time across Qiskit, Cirq and Braket backends.
Use it when
Reach for it when you want a vendor-neutral, cross-platform comparison you can rerun yourself.
Background: why quantum benchmarking became policy
Benchmarking & characterization
Sandia National Laboratories' toolkit for characterizing noisy processors. Its flagship protocol is gate set tomography, alongside randomized benchmarking and robust phase estimation, with recent work on leakage-aware models.
Use it when
Use it when you need to know precisely how a device's gates are wrong, not just how often results are.
Background: quantum volume
Languages & intermediate representations
The de facto interchange language for quantum circuits. Version 3 adds classical control flow, timing and pulse-level constructs, and is now governed by an independent technical steering committee rather than a single vendor.
Use it when
Reach for it when a circuit has to move between two toolchains that share nothing else.
Languages & intermediate representations
An LLVM-IR-based representation for quantum programs, letting front ends such as Q# and Qiskit target many backends through one compiler layer. The alliance also ships `qir-runner`, a Rust sparse-state simulator, and PyQIR for generating QIR from Python.
Use it when
Use it when you are building the toolchain rather than the algorithm — a compiler, a backend, or anything that has to consume programs it did not author.
Languages & intermediate representations
Microsoft's standalone quantum language and toolchain: compiler, standard library, local simulators with noise, Jupyter kernel and the resource estimator. Q# treats qubits as a managed resource with explicit classical/quantum separation rather than as a Python library.
Use it when
Reach for it when you want a real type-checked language and a debugger rather than building circuits by appending to a list.
Languages & intermediate representations
An ETH Zurich research language whose headline feature is automatic uncomputation: temporary values are safely discarded by the type system, so you write far less bookkeeping than in circuit-level languages. It ships a type checker, compiler and interpreter.
Use it when
Worth a look when you want to see where high-level quantum language design is heading — it is a research project, not a production toolchain.
Editor & IDE tooling
The most complete quantum editor experience available: IntelliSense and debugging for Q# and OpenQASM, local noisy simulators, circuit visualization, resource estimation and job submission — all inside VS Code, with a browser version at vscode.dev.
Use it when
Use it when you want to step through quantum code in a debugger instead of inferring what happened from printed counts.
Resource estimation
A Google Quantum AI library for expressing fault-tolerant algorithms as composable 'bloqs' and costing them out in qubits, T gates and magic states. Maintained by Google but self-described as an experimental preview.
Use it when
Reach for it when the question is not 'does my algorithm work' but 'how large a fault-tolerant machine would it actually need'.
Background: logical qubits and fault tolerance explained
Resource estimation
Estimates the physical qubit count, runtime and error-correction overhead a program would require under configurable qubit and QEC-code assumptions. It runs locally in VS Code with no Azure account, and is free in every access path.
Use it when
Use it when you need a defensible number for how far away a fault-tolerant version of your workload is.
Background: what a logical qubit costs
Physics & domain frameworks
The standard Python library for simulating the dynamics of open and closed quantum systems — Lindblad master equations, Monte Carlo trajectories, time-dependent Hamiltonians — rather than idealized gate circuits.
Use it when
Reach for it when you care about the physics underneath the gates: decoherence, pulse shapes, or a Hamiltonian that never becomes a circuit.
Physics & domain frameworks
A full-stack open framework covering circuit definition, hardware-agnostic CPU/GPU/multi-GPU backends and a built-in algorithm library — plus sibling projects Qibolab and Qibocal that drive and calibrate real control electronics.
Use it when
Use it when you are working on the hardware side and need one stack that spans algorithms down to pulse-level calibration.
Background: free quantum hardware access
Physics & domain frameworks
Quandela's framework for photonic quantum computing: linear-optics circuit construction, single-photon source error models, Fock-state manipulation, simulation backends and job submission to Quandela's cloud.
Use it when
Reach for it when your problem is photonic — gate-model SDKs assume a qubit model that does not describe linear optics well.
Background: free quantum hardware access
The five mainstream SDKs are deliberately absent — Qiskit, Cirq, PennyLane, Braket and CUDA-Q each have their own setup guide under SDKs, including Qiskit, PennyLane and CUDA-Q. This page is about what you reach for once an SDK alone is not enough.
We also dropped several things we could not stand behind. The Qiskit Code Assistant VS Code extension was archived in May 2026 after IBM discontinued the service behind it, so there is no live official Qiskit editor extension to recommend. The Intel Quantum Simulator is not archived, but its only recent commits are automated dependency bumps and it has never cut a release — we would not build on it. And cnot.io, which we list elsewhere as reading material, publishes no author, licence or changelog, so it does not belong in a tools list.
Two entries here carry warnings rather than exclusions. Quirk has not seen a commit since 2022, but the hosted app still works and nothing has replaced it for quick exploration. qiskit-qec is listed only so you know it exists and is stalled — Stim and PyMatching are the maintained path.
Tools make more sense once you know which problem they are solving.
What developers actually ask once they move past their first circuit.
Almost all of them. Every tool on this page is free to use, and the large majority are open source under permissive licences such as Apache-2.0, MIT or BSD. The two exceptions are IBM Quantum Composer, a free web tool that is not open source, and NVIDIA's cuQuantum, whose core libraries are proprietary but free to download — only its Python bindings are BSD-licensed.
An SDK such as Qiskit or Cirq gives you a way to express and run a circuit end to end. The tools here each solve one problem better than a general SDK can: TKET and BQSKit compile circuits harder, Mitiq corrects noise-induced bias, Stim and PyMatching handle error correction at scales state-vector simulators cannot reach, and Metriq or SupermarQ tell you whether a device is as good as its vendor claims.
In practice, Stim plus PyMatching. Stim simulates stabilizer circuits fast enough for surface-code experiments and emits a detector error model; PyMatching decodes the resulting syndromes with minimum-weight perfect matching. Qualtran or the Azure Quantum Resource Estimator then tell you what the fault-tolerant version would physically cost.
Not one you should rely on in 2026. IBM archived the Qiskit Code Assistant extension in May 2026 after discontinuing the service behind it. The best-supported quantum editor experience today is Microsoft's Azure Quantum Development Kit extension, which covers Q# and OpenQASM with IntelliSense, debugging, local noisy simulation and resource estimation.
Check the repository's latest commit and its most recent tagged release, and treat those as separate signals — a repo can look busy while only receiving automated dependency bumps. Every entry on this page records a status based on exactly that check, but statuses go stale, so verify before you build something long-lived on a project.