European superconducting quantum hardware provider IQM Quantum Computers and NVIDIA have demonstrated the agentic AI calibration of a 20-qubit IQM superconducting quantum processor using specialized NVIDIA Ising Calibration vision-language models (VLMs).
Integrated directly into IQM’s proprietary hardware control and calibration software, the agentic VLM stack analyzed raw experimental tuning charts, autonomously determined optimal parameter adjustments, and executed dynamic bring-up routines across all 20 physical qubits. The workflow automatically identified out-of-bounds qubit parameter settings, dynamically adjusted measurement parameter ranges, and generated plain-English diagnostic logs explaining each calibration decision for human auditability.
| [ IQM / NVIDIA QCalEval Open Calibration Benchmark & VLM Architecture ] | ||
|---|---|---|
| System Module | Technical Implementation & Data Scope | Performance Metric & Operational Function |
| • AI Calibration Engine | • NVIDIA Ising Vision-Language Models (VLMs) | • Open model family fine-tuned for quantum calibration chart interpretation and parameter decisioning. |
| • Deployment Footprint | • Compressed edge VLM deployment on NVIDIA DGX Station | • Runs locally at supercomputing centers and QPU cryogenic labs without data center dependencies. |
| • QCalEval Benchmark | • Open benchmark with 243 charts across 22 experiment types | • Scores model diagnostic accuracy across 6 standardized diagnostic questions per chart. |
| • Hardware Validation | • 20-Qubit IQM Superconducting QPU Stack | • Autonomous bring-up achieved; NVIDIA Ising scored up to 19 points higher than general-purpose VLMs. |
Standardizing Agentic QPU Bring-up with the Open QCalEval Benchmark
To establish a standardized metric for AI-driven quantum processor bring-up, NVIDIA and IQM co-authored and released QCalEval—an open benchmark dataset featuring 243 calibration charts across 22 distinct experiment types (sourced from real superconducting hardware and simulated neutral-atom systems). Each chart presents six diagnostic questions evaluating a model’s ability to identify parameter anomalies and select corrective actions.
On the public QCalEval leaderboard, specialized NVIDIA Ising Calibration models scored up to 19 points higher than general-purpose vision-language models. The open-weights architecture allows quantum laboratories and High-Performance Computing (HPC) centers to deploy compressed Ising models locally on edge devices such as the NVIDIA DGX Station, establishing a scalable, human-auditable framework to eliminate manual engineering bottlenecks as superconducting QPUs scale past hundreds of qubits.
Review the technical announcement on the IQM Blog here.
In GQI Portal
The players behind the news
The team that writes QCR tracks every company, deal and technology in the GQI Factory, GQI's verified database of the quantum industry. Next up: every story linked to its players, coming to QCR's paid plans.
- Players Companies and institutions across the quantum industry, by segment.
- Scorecards How the players compare on hardware, software, funding and more.
Newsletter
QCR Alerts in your inbox
The latest reporting and analysis from Quantum Computing Report. Free, unsubscribe any time.