Original data

Most popular quantum computing frameworks

All 22 major quantum programming frameworks, ranked by GitHub stars, then ranked again by a year of actual upkeep. The two rankings disagree sharply, and the gap is the most useful thing on this page.

By Dr. Donovan. Live data from the GitHub API, as of 24 July 2026. Refreshed with scripts/fetch_framework_stars.py and scripts/fetch_framework_activity.py.

What stars do and do not mean Stars measure community interest and visibility, not quality or fitness. Vendor-backed projects (IBM, Google, NVIDIA, Microsoft) draw more stars. Read alongside the gotchas and your actual use case.

Framework, library, or language?

People say "quantum programming language" for all of these, but most are not languages. A library (or framework, or SDK) is code you import into a language you already know: you write ordinary Python and call Qiskit, Cirq, or PennyLane. A language has its own syntax and compiler that you learn from scratch.

Of the 22 here, only Q# and OpenQASM are genuine languages: Q# is Microsoft's dedicated quantum language, with its own syntax you learn from scratch, and OpenQASM is a low-level, assembly-style format that other tools compile to rather than something you write whole programs in. Everything else is a library you call from a language you already know, most of them from Python, a couple from Julia (Yao, Bloqade), and one, tket, whose fast C++ core you drive from Python.

One more distinction the table makes explicit: the language a project is built in is not the language you write in. tket is written in C++ but you use it from Python; Q#'s compiler is written in Rust but you write Q#. The "you write in" column is the one that matters when you sit down to code.

Ranked by GitHub stars

# Framework Stars Forks Since Activity You write in License
1 Qiskit 7,638 2,982 2017 · 9.4y active Python Apache-2.0
2 Cirq 5,020 1,249 2017 · 8.6y active Python Apache-2.0
3 PennyLane 3,367 840 2018 · 8.3y active Python Apache-2.0
4 TensorFlow Quantum 2,172 658 2020 · 6.5y active Python Apache-2.0
5 QuTiP 2,046 776 2012 · 13.8y active Python BSD-3-Clause
6 OpenFermion 1,714 430 2017 · 8.8y active Python Apache-2.0
7 PyQuil 1,494 358 2017 · 9.5y active Python Apache-2.0
8 OpenQASM language 1,488 359 2017 · 9.4y active OpenQASM Apache-2.0
9 CUDA-Q 1,100 426 2023 · 3.4y active Python or C++ Apache-2.0
10 Yao.jl 1,037 131 2018 · 8.3y slowing Julia other
11 Q# language 974 205 2023 · 3.5y active Q# MIT
12 Strawberry Fields 853 224 2018 · 8.3y stale Python Apache-2.0
13 Stim 789 221 2021 · 5.4y active Python or C++ Apache-2.0
14 D-Wave Ocean 531 188 2017 · 8.7y active Python Apache-2.0
15 Mitiq 441 233 2020 · 6.5y active Python GPL-3.0
16 Amazon Braket SDK 371 192 2019 · 6.7y active Python Apache-2.0
17 Qibo 360 96 2020 · 6.4y active Python Apache-2.0
18 tket (pytket) 314 61 2021 · 4.9y active Python Apache-2.0
19 Bloqade.jl 212 44 2021 · 4.8y slowing Julia other
20 Perceval 203 91 2022 · 4.3y active Python other
21 Superstaq 129 24 2022 · 4.4y active Python Apache-2.0
22 OpenQL 109 51 2017 · 9.3y stale Python or C++ other

Stars, forks and dates from the GitHub API on 24 July 2026. "Activity" is time since the last commit: active (within 30 days), slowing (within 6 months), stale (longer). "Since" is the year the repository was created.

Popular is not the same as maintained

Stars accumulate and never expire. A project that was exciting in 2019 keeps its stars even if nobody has touched it since, which is why a ranking by popularity quietly flatters abandoned code. So we measured the other thing: for each repository, every week of commit activity over the last 52 weeks, every release shipped in the last 12 months, and the total number of people who have ever landed a commit.

Ranking the same 22 projects by commits over the last year instead of stars reorders the list substantially. Qibo moves up 15 places, from 17th by stars to 2nd by commits. In the other direction, PyQuil and TensorFlow Quantum are both in the top 7 by stars and near the bottom by work done.

# Framework vs stars Commits52 weeks Weekly pattern Activeweeks Upkeep People Releases12 months Latest
1 PennyLane ▲2 1,419 52/52 steady 216 11 Jun 2026
2 Qibo ▲15 1,379 49/52 steady 59 8 Jun 2026
3 Qiskit ▼2 1,195 51/52 steady 462 22 Jul 2026
4 CUDA-Q ▲5 987 51/52 steady 104 5 Jul 2026
5 QuTiP same 675 48/52 steady 190 5 May 2026
6 Q# ▲5 545 51/52 steady 64 11 Jul 2026
7 Cirq ▼5 403 51/52 steady 263 2 Jun 2026
8 D-Wave Ocean ▲6 271 44/52 steady 22 5 Jun 2026
9 Amazon Braket SDK ▲7 243 44/52 steady 75 60 Jul 2026
10 OpenFermion ▼4 188 33/52 intermittent 87 2 Jul 2026
11 Mitiq ▲4 157 37/52 intermittent 93 5 Mar 2026
12 tket (pytket) ▲6 145 42/52 steady 36 21 Jul 2026
13 Perceval ▲7 136 40/52 steady 39 7 Jul 2026
14 Superstaq ▲7 121 43/52 steady 38 19 May 2026
15 TensorFlow Quantum ▼11 102 27/52 intermittent 43 2 Feb 2026
16 OpenQASM ▼8 76 26/52 intermittent 78 – May 2024
17 Stim ▼4 57 20/52 intermittent 41 1 May 2026
18 Yao.jl ▼8 20 11/52 dormant 34 2 Mar 2026
19 PyQuil ▼12 11 4/52 dormant 96 10 Jul 2026
20 Bloqade.jl ▼1 3 2/52 dormant 25 – Apr 2025
21 Strawberry Fields ▼9 2 2/52 dormant 43 – Jun 2022
22 OpenQL same 0 no commits 0/52 dormant 23 – Jan 2024

Commit counts, releases and contributors from the GitHub API on 24 July 2026, covering the 52 weeks to that date. "vs stars" is the movement between this ranking and the star ranking above. "Active weeks" counts weeks with at least one commit, which separates steady work from a single burst. Each sparkline is scaled to that project's own busiest week, so compare the shape rather than the height; the busiest single week in the whole set was 96 commits. "People" is everyone who has ever landed a commit, not current maintainers.

What this changes if you are picking one to learn 5 of the 22 were quiet for most of the last year: OpenQL (0 active weeks, 0 commits), Bloqade.jl (2 active weeks, 3 commits), Strawberry Fields (2 active weeks, 2 commits), PyQuil (4 active weeks, 11 commits), Yao.jl (11 active weeks, 20 commits). Several still appear on published "top quantum frameworks" lists. If you are choosing where to spend a few months of learning, that is the column to read first, because a dormant project means unanswered issues, unfixed breakage against new Python versions, and tutorials that quietly stop working. 12 of the 22 committed in 40 or more of the last 52 weeks.

What the numbers say

The 22 projects hold 32,362 stars and 9,839 forks between them, but the distribution is top-heavy: Qiskit, Cirq and PennyLane alone are 50% of all the stars. The head of the field is the IBM, Google and Xanadu triad, exactly the projects with the largest teams and marketing behind them, which is the first reason to read stars as visibility rather than merit.

Python owns the ecosystem: 15 of 22 projects are built in Python, and nearly everyone writes Python even when the engine underneath is C++ (Stim, tket) or Rust (Q#'s compiler). Julia (Yao, Bloqade) is the only real alternative host language with traction.

Adjusted for age, the fastest growers are Qiskit (~813 stars/year), Cirq (~583 stars/year), PennyLane (~407 stars/year), TensorFlow Quantum (~336 stars/year), with NVIDIA's CUDA-Q and Microsoft's Q# standing out as the young projects climbing quickest. Maintenance is the counterweight, and it is measured properly in the section above: 18 of 22 projects were touched within the last month, but a single commit is a low bar, and counting whole weeks of activity across a year separates the genuinely staffed projects from the ones receiving occasional dependency bumps.

One gotcha for each

Popularity does not tell you what will bite you. Here is the single thing to know about each, in the same star order.

Qiskit
Big API breaks between major versions; older tutorial code often fails on 1.x.
Cirq
Qubits carry hardware placement; fewer high-level algorithm helpers than Qiskit.
PennyLane
Built around autodiff; overkill if you are not doing QML, and speed depends on the backend.
TensorFlow Quantum
Pinned to specific TensorFlow and Cirq versions; fragile to install and slow to update.
QuTiP
A dynamics and open-quantum-systems simulator, not a gate-circuit framework for hardware.
OpenFermion
A chemistry/fermion operator library used with Cirq or others, not a standalone framework.
PyQuil
Barely developed now, quiet for most of the last year; needs external quilc and QVM binaries, and is Rigetti-specific.
OpenQASM
A low-level assembly-style language and interchange format, not one you write whole programs in.
CUDA-Q
Shows its value on NVIDIA GPUs; younger, fast-moving API and heavier setup.
Yao.jl
Development has slowed sharply, with commits in only a minority of recent weeks; Julia-only, so a smaller ecosystem, first-run compile latency, and fewer tutorials.
Q#
A separate language, not a Python library, so a bigger learning curve; hardware runs via Azure.
Strawberry Fields
Photonic continuous-variable model, not qubit gates; effectively in maintenance mode.
Stim
Blazing fast but stabilizer/Clifford only; not a general universal simulator.
D-Wave Ocean
Quantum annealing only; you must recast problems as QUBO/Ising, and it targets D-Wave hardware.
Mitiq
An error-mitigation toolkit you bolt onto Qiskit or Cirq, not a circuit framework itself.
Amazon Braket SDK
A thin AWS cloud client; hardware runs cost money and you work inside the AWS ecosystem.
Qibo
Smaller, academic-led community; thinner ecosystem and docs than the majors.
tket (pytket)
Primarily a circuit compiler and optimizer, not where you author circuits; C++ core.
Bloqade.jl
Julia-only and specific to neutral-atom analog (QuEra) hardware; niche, lower activity.
Perceval
Photonic linear-optics specific (Quandela); a different paradigm with a small ecosystem.
Superstaq
A cross-compiler cloud-service client (Infleqtion), not a full framework; ties to their service.
OpenQL
Low activity (no commits in roughly two years); academic and effectively stale.
Next step Stars pick the crowd's favourite, not yours. To match a framework to what you are actually trying to do, use the framework picker and comparison, or read the head-to-head on Qiskit vs PennyLane.

Method and sources

Every number here comes from the public GitHub REST API, read on 24 July 2026. Popularity figures come from the repos endpoint via scripts/fetch_framework_stars.py. Maintenance figures come from stats/commit_activity, releases and contributors via scripts/fetch_framework_activity.py. Both are re-runnable, so the figures are refreshed on demand rather than drifting out of date, and each project links to the repository it was measured from. The editorial fields, what you write in, whether a tool is a language, and the gotchas, are hand-authored and reviewed.

Two limits worth stating. Commit counts reward projects that merge many small commits and under-credit those that squash, so compare the number of active weeks as well as the raw total. Contributor counts include anyone who has ever landed a commit, not just current maintainers. And stars are a proxy for community interest and visibility, not a measure of quality, performance, or fitness for your task: a small, well-run project serving a niche will never out-star a vendor-backed one, and nothing in either table should be read as a quality score. The tables answer "how much attention does this have" and "is anyone still working on it". They do not answer "is this good", which depends entirely on what you are building.