IBM has launched Qiskit SDK v2.0, a major update that introduces foundational changes to the quantum software stack, including the first version of a C-language API, deeper integration of Rust-based internals, and removal of legacy components. The release aligns with Qiskit’s 18-month major version support cycle and reflects IBM’s strategic effort to enhance performance, improve interoperability with high-performance computing (HPC), and simplify the development of quantum-classical hybrid applications.
A key feature of v2.0 is the new C API for constructing and manipulating SparseObservable objects. This early-stage interface allows C programs to interface directly with Qiskit data structures and is part of a longer-term roadmap to support large-scale quantum experiments beyond Python environments. The integration is enabled by Qiskit’s ongoing backend transition to Rust, which now underpins several core data structures including qubit registers and instruction classes. This architecture is designed to facilitate low-level access and interoperability with classical infrastructure.
Performance improvements continue to be a major focus. Circuit construction tasks see a 2x speedup, and transpilation benchmarks indicate an average 20% performance gain compared to Qiskit v1.3, with improvements as high as 17x for specific workloads. However, benchmarks using OpenQASM 2 classically conditioned gates show a runtime increase in 11% of tests due to more accurate modeling using IfElseOp constructs. These tradeoffs reflect an effort to unify semantic correctness with performance optimization as Qiskit moves toward full OpenQASM 3 compatibility.
The update also introduces stretch variables for Delay instructions and a new box instruction for grouping operations. Stretch variables allow abstract expression of timing dependencies in dynamic circuits, aiding the design of error suppression methods such as dynamical decoupling. The box instruction provides compiler-visible logical grouping of gates for use cases like Pauli twirling, noise learning, and scheduling control. Both features align with OpenQASM 3 semantics and prepare Qiskit for more advanced compilation and runtime behaviors on IBM hardware.
As part of breaking changes, Qiskit SDK v2.0 removes several deprecated components including qiskit.pulse, qobj, and BackendV1. These removals simplify the SDK and reflect IBM’s shift away from legacy control models in favor of higher-level primitives and fractional gates for low-level control. IBM encourages former Qiskit Pulse users to explore alternatives such as Qiskit Dynamics for calibration and optimal control. Full release notes, migration guidance, and roadmap updates are available on the Qiskit documentation site.
You can find the Qiskit SDK v2.0 release summary here, the migration guide here, the release notes here, installation instructions here, IBM Quantum learning courses here, updated guides and API docs here, new C API documentation here, the Qiskit 2.x roadmap here, and the GitHub issue tracker here.
April 4, 2025
