Quantum software developer Quantum X Labs Inc. (Nasdaq: QXL) has announced new performance results from its AI-driven quantum error correction (QEC) decoder program. Testing its updated model against Google’s public surface-code experimental dataset, Quantum X Labs demonstrated improved decoding accuracy compared to standard matching-family baselines—including Google’s published correlated-matching and PyMatching benchmark results for the same surface-code configuration.
[ Quantum X Labs AI-QEC Decoder Architecture ]
│
┌──────────────────────────────────┴──────────────────────────────────┐
▼ ▼
Synthetic AI Training Pipeline Real-Hardware Syndrome Generalization
• Trained Exclusively on Synthetic Samples. • Tested on Google Surface-Code Experimental Dataset.
• Syndrome Info & Error Weighting Integration. • Outperforms PyMatching & Correlated-Matching.
• NVIDIA CUDA-Q & GPU Acceleration. • Low-Latency Foundation for Real-Time QEC.Synthetic-to-Real Generalization and GPU Acceleration
A central challenge in real-time quantum error correction is developing decoders that can interpret physical syndrome data rapidly without incurring prohibitive computational latency or requiring extensive retraining on real hardware shots:
- Zero-Shot Real Hardware Generalization: QXL’s updated AI decoder model was trained exclusively on synthetic simulation samples and was not exposed to real hardware shots during training. Achieving higher accuracy than standard minimum-weight perfect matching (MWPM) solvers on real experimental data confirms the model’s ability to generalize to physical device noise.
- GPU-Accelerated Integration: Built to leverage NVIDIA CUDA-Q and accelerated GPU computing architectures, the decoder combines surface-code topological structures with AI-based error weighting to provide a low-latency path toward real-time decoding on fault-tolerant systems.
- Next Steps on Roadmap: Led by Chief Quantum Technology Scientist Prof. Nir Sharon, Quantum X Labs plans to extend and replicate the synthetic-to-real decoding pipeline across additional physical hardware backends, code topologies, and device centers.
The software milestone advances commercial QEC decoding workflows required to scale surface-code quantum computing toward fault-tolerant operation.
Review the announcement on GlobeNewswire here.
August 21, 2026