Quantum Powerhouse

The field · tooling

Open-source tooling

The stack behind the software-engineer roles. This site's own tools sit inside it: the transpiler speaks OpenQASM and targets Braket, the Circuit Lab exports OpenQASM 2.0, and pytest-qequiv asserts Qiskit, Cirq and Braket circuits agree. Licenses and dates below come from the repositories themselves.

  • Verified — published result
  • Vendor-reported result
  • Projection — roadmap target
  • Opinion — named, dated
  • Estimate — with caveats
  • Contested — disputed in the record
  • Preprint — not yet peer-reviewed

The frameworks

Qiskit SDK 2.0 (IBM) — Apache-2.0

Verified — published result

Qiskit 2.0.0 was released on 31 March 2025. The release removed legacy APIs (the .c_if() method, the qobj module, BackendV1 and the Pulse package), added a C API for sparse observables — the SDK's first public C interface — and reported a 2× speedup in circuit construction over v1.3 from its Rust core. Qiskit is the SDK the site's own transpiler pipeline targets; the research wing's Benchpress entry (claim C14) is IBM's pytest-based cross-SDK benchmark built around it.

OpenQASM 3.1.0 — the circuit interchange language

Verified — published result

The OpenQASM specification's latest release is v3.1.0 (15 May 2024). OpenQASM 3 is the intermediate representation higher-level compilers use to talk to hardware, adding classical feed-forward control flow based on measurement outcomes and timing constructs (including 'stretch' for deferred timing resolution). The transpiler on this site parses OpenQASM 2.0 and 3.0; the Circuit Lab exports 2.0.

CUDA-Q (NVIDIA) — hybrid CPU/GPU/QPU programming, Apache-2.0

Verified — published result

CUDA-Q is NVIDIA's open-source (Apache-2.0) platform for hybrid quantum-classical programs in C++ and Python, with a unified model for CPUs, GPUs and QPUs working together — the toolchain for the hybrid-systems engineer role in the careers section.

Stim (Google) — the error-correction simulator

Verified — published result

Gidney's Stim simulates stabilizer circuits using an inverse-tableau representation, SIMD-optimized data layouts and Pauli-frame propagation. The paper reports analyzing a distance-100 surface-code circuit (20,000 qubits, 8 million gates, 1 million measurements) in 15 seconds and then sampling full shots at 1 kHz. It is the workhorse behind the below-threshold experiments on the hardware scoreboard.

Ecosystem snapshot: licenses and adoption (GitHub, August 2026)

Estimate — with caveats

All of the major frameworks are Apache-2.0 licensed. GitHub stars on the check date: Qiskit 7,731; Cirq (Google) 5,052; PennyLane (Xanadu, automatic differentiation for hybrid quantum-classical programs) 3,431; CUDA-Q 1,118; Stim 811; the OpenQASM specification 1,500. Stars measure attention, not quality — a dated proxy, nothing more.

See also