July 2026 (v2026.07)

• QruiseML
• QruiseOS
  • New features
  • Improvements
  • Bug fixes

QruiseML

This release marks the launch of v3 of our simulation and modelling library, qruise-toolset. Designed from the ground up, this version uses a high-performance and memory-efficient Julia-based simulation engine while retaining the familiar user-facing Python API.

As the community moves from the NISQ to the FTQC era, there's increasing demand for modelling physically accurate system dynamics at the scale of standard QEC unit cells, i.e. 15–20 qubits. At the same time, an automatically differentiable simulator is needed for gate optimisation and parameter-learning workflows. Meeting these demands required a complete rethink of the software architecture. You can see below a preliminary benchmark comparing the performance of qruise-toolset@v3 with other libraries in the ecosystem.

QruiseOS

This focuses on enhancing the user dashboard experience, with new features and improvements across all main pages.

New features

Branch comparison and experiment links in System Info

The System Info page now has the option to compare two experiment branches side-by-side, allowing you to easily compare different qubit metrics across branches. (Branching itself already existed; this release adds the side-by-side comparison view and extends branch tracking to flow runs, not just experiments.)

Qubit metrics now include links to the corresponding experiment in the database.

Summary table for experiments

When viewing an experiment in either workflow view or the Experiment Database, you can now see a summary table of the metrics derived from that experiment.

QPU metric history

A new history view on the System Info page lets you track how your QPU metrics have evolved over time. You can select the maximum number of commits to view (history limit), the time interval, and whether the metric is plotted against the date or commit number. There's also an option to overlay the data from multiple qubits and branches for easier comparison.

Deployments in the dashboard

You can now run, edit, and schedule deployments in the dashboard, making it easier to manage deployments from a single interface.

Richer experiment data downloads

Experiment data exports now include a provenance.yaml file containing experiment, software, and version information, together with a snapshot of the QPU state (kb_snapshot.json) after the experiment completed. This makes it easier to reproduce and analyse results.

Improvements

More compact Experiment Database

The Experiment Database has been redesigned with a more compact layout, sticky checkboxes, and a consolidated filter bar, making it easier to filter, compare, and export experiments. An additional compact view is also available for displaying more experiments on smaller screens.

Flow run branch visibility

The flow run branch is now shown wherever flow runs are listed. Previously this was only shown in the Experiment Database.

Consolidated experiment addressing

The API and UI now use a single, consistent path to address experiments by name.

Bug Fixes

• Fixed a bug where users with no permitted QPUs could, in some cases, see runs and workflows belonging to other QPUs.
• Fixed intermittent JupyterHub connection errors.