Product · Optimisation

Exact QUBO optimisation, guaranteed global optimum.

A GPU-accelerated branch-and-bound engine that returns the proven minimum of any Quadratic Unconstrained Binary Optimisation problem mapped onto D-Wave's Chimera graph — up to 1024 variables, every call.

Submit through a Python SDK that mirrors the dimod interface and use the result as ground truth for molecular optimisation, logistics, portfolio selection, and network design.

Exact
GPU
dimod-compatible
1024 vars
Problem size
≤ 1024 vars

Solved exactly on the 1024-qubit Chimera topology — the same footprint quantum annealers target, with provable certainty instead of probabilistic samples.

Per request
~$0.30

Pay-per-call cloud delivery keeps the cost of a proven optimum low enough for benchmarking heuristics in tight feedback loops.

Capabilities

Exact algorithms, GPU parallelism, drop-in API — a ground-truth reference for QUBOs that matter.

F.01

Patented branch-and-bound

A patented exact algorithm prunes the search tree aggressively, returning the proven global optimum rather than a best-effort approximation.

F.02

GPU-parallel search

Bound evaluation and subtree expansion run in parallel on cloud GPUs, collapsing what would be hours of CPU search into seconds.

F.03

dimod-compatible API

A Python SDK that mirrors D-Wave's dimod interface — drop the solver into existing optimisation pipelines without rewriting your problem model.

F.04

Chimera-graph alignment

Tuned for QUBOs mapped onto the 1024-qubit Chimera topology, giving the same problem footprint as a 2000Q annealer with classical certainty.

F.05

Deterministic results

The same input returns the same proven optimum every run — a ground-truth reference for benchmarking heuristics and validating quantum samplers.

F.06

Cloud-native delivery

Pay-per-call cloud service, no hardware to provision. Submit a problem, retrieve the optimum, move on with your pipeline.

Workflow

How a QUBO Solver run reaches the proven optimum.

  1. 01Formulate the QUBO as a coefficient matrix or dimod BQM
  2. 02Submit through the Python SDK to the cloud endpoint
  3. 03Evaluate bounds in parallel across GPU cores
  4. 04Prune the search tree with patented branch-and-bound
  5. 05Return the proven minimum and its assignment
  6. 06Use as ground truth for downstream pipelines