Q-Prime
Quantum embeddings
The foundation layer of the QGI stack — encodes polarity, scope, conditions, and rule dependencies as structure, not similarity scores.
Distributed under the QGI Commercial Model License v1.0. No QPU required — runs on NVIDIA CUDA-Q.
What Q-Prime does differently
Test today on HuggingFace — compare against your current embedding model.
Quantum-structured hypergraph
Encodes guidelines, contracts, and policies as a hypergraph preserving polarity, scope, conditions, and cross-rule dependencies.
Commercial model, managed API
On HuggingFace under the QGI Commercial Model License v1.0. Weights are not shipped — access via managed API.
No QPU required
Runs on NVIDIA CUDA-Q and cuTensorNet on commodity GPUs. Quantum is the representation, not the hardware.
Deterministic inference
Same input, same structural encoding every time — what regulators expect for replay.
Regulatory-grade provenance
Every embedding carries rule ids, source documents, and condition graph for audit examination.
Trained on regulated corpora
Pre-trained on compliance, contract, and lending corpora — calibrated for regulated decisions.
Q-Prime is the foundation. The QAG Engine is the reasoning.
Q-Prime encodes structure. The QAG Engine reasons over it via seven interpretable HSC signals.
Questions developers ask
How do I get access to Q-Prime?
Request access on HuggingFace at QGI-dev/q-prime under the QGI Commercial Model License v1.0.
Can I self-host the weights?
Not today — managed API only. Self-hosting is available under enterprise engagement.
Does Q-Prime work without the QAG Engine?
Yes, as an encoding layer. Pair with the QAG Engine to unlock the seven HSC signals.
What hardware do I need?
NVIDIA GPUs with CUDA-Q / cuTensorNet. No quantum hardware required.
Ready to encode your regulated corpus as structure?
Request access on HuggingFace or talk to us about a demo.