Picture a materials scientist we’ll call Priya. She is an illustrative example, not a real person. She works in a university lab and has spent years writing simulations in VASP, a chemistry code with deep roots in Fortran. A colleague suggests she test a quantum circuit alongside her model. Her heart sinks, because the easy route runs through Python, and rewriting her workflow around it could eat weeks before a grant deadline.
New Qiskit C API bindings from IBM aim at exactly that problem. IBM published its blog post on October 6. The bindings connect Qiskit, its open source quantum software, to Fortran, C++, and Julia, so researchers can call a quantum circuit from inside their existing code instead of rebuilding the project in Python first.
What the bindings do
Qiskit is IBM’s toolkit for programming quantum computers, and most people use it through Python. But many chemistry, physics, and materials codes run on compiled languages like Fortran and C++. The new tools are named qiskit-fortran, qiskit-cpp, and Qiskit.jl. They sit on Qiskit’s C API, which has existed since Qiskit v2.0. IBM says calls from these tools reach the same Rust core that powers the Python SDK, “without going through Python at any point.”
IBM lists existing codes where this could help, including GAMESS, Quantum ESPRESSO, CP2K, and VASP for chemistry and materials work, plus LAMMPS and GROMACS for physics simulation. It frames the payoff as “tighter, more performant integration” with existing Fortran applications. The Quantum Computing Report notes that all three bindings share one underlying library, so a circuit built in Fortran can pass to a C++ or Julia module, and says the tools avoid Python’s Global Interpreter Lock overhead.
| Language | Tool | How it connects | Source |
|---|---|---|---|
| Fortran | qiskit-fortran | Uses Fortran’s iso_c_binding to build and edit circuits | IBM Quantum blog |
| C++ | qiskit-cpp | Header-only interface to the C API | IBM Quantum blog |
| Julia | Qiskit.jl | Wraps the C API in native Julia types, with no build from source | IBM Quantum blog |
Works well, and be careful
Teams who want to test a quantum idea inside an existing HPC project usually export data out of the Fortran or C++ code, convert it for Python, write and debug a wrapper, run the quantum part separately, and bring the results back. Each step takes time and adds room for bugs. IBM and the Quantum Computing Report do not put a number on the time saved, and we won’t guess one.
| Works well | Be careful |
|---|---|
| Calls reach Qiskit’s Rust core without going through Python, per IBM. | IBM says all of these non-Python bindings are “in active development.” It does not label any of them as stable. |
| All three bindings share one library, so circuits can pass between Fortran, C++, and Julia modules. | Neither IBM nor the Quantum Computing Report publishes benchmarks. The speed and interoperability claims are not quantified. |
| IBM’s demo ran a 100-qubit transverse-field Ising model through mapping, transpiling, running, and post-processing. | The 100-qubit demo cannot be checked exactly: a full state-vector solution needs roughly 10^30 amplitudes. The tensor-network reference is reliable only while the state is weakly entangled, typically early in the run. |
| At 20 qubits, IBM checked the output against an exact solution and a noiseless tensor-network simulation. | Smaller Trotter steps cut discretization error but deepen circuits and add hardware noise, so accuracy stops improving past a point. |
| A second Julia package, QiskitIBMRuntime.jl, submits circuits to IBM Quantum hardware. | qiskit-cpp is not brand new; IBM says it was first announced the previous year. The Fortran and Julia tools are the newer pieces. No independent lab has reported results yet. |
Bottom line
The demo is a software demo, not a scientific discovery. It shows the tools connect. It doesn’t show that a Fortran lab gets a better result than it would with Python. This release won’t set a flashy qubit record; it clears a practical path for researchers who already work in Fortran or C++. The same kind of groundwork is showing up elsewhere, including Oak Ridge’s LuGo algorithm, which reports a 95 percent cut in gate count for one fluid-flow problem. Watch for the first published papers that cite these bindings. That is when we’ll learn whether the shortcut changed what gets discovered, not just how the code is wired.
Why It Matters to You
If you write research code in Fortran, C++, or Julia, you can try quantum circuits inside your existing project without a Python detour, which matters when grants run on one- or two-year clocks and glue code can crowd out science. The tools are early, though: IBM calls them in active development and invites feedback through issues and pull requests. Anyone outside research computing won’t notice a direct change.
Quick answers
Do I still need Python to use Qiskit? Not for these tools. IBM says calls from the new bindings reach Qiskit’s Rust core without going through Python. Python remains the main way most people use Qiskit.
Are the bindings ready for production? Treat them as early tools. IBM describes them as in active development.
Will this save my lab money? Neither IBM nor the Quantum Computing Report gives a cost or time figure. Any savings depend on how much glue code your team would otherwise write.
Sources
- IBM Quantum Blog: Qiskit Beyond Python (Oct 06, 2026)
- Quantum Computing Report: IBM Expands Qiskit Beyond Python with Native C API Bindings (Oct 10, 2026)
