v0.4.0 Available 832 Tests Collected Git 729282f

Quantum Error Correction
Decoding, Reproducible by Design

A source-available Rust/Python QEC decoder platform with Stim/Sinter-compatible workflows, PyMatching-compatible interfaces, belief-matching accuracy mode, BP-OSD for LDPC/qLDPC codes, CPU/GPU batch decoding, and artifact-hashed reproducible benchmarks.

View QECTOR Decoder Request Commercial License Workbench — Coming Soon

Free for personal, academic, educational, and non-commercial research use. Commercial use requires a paid license. Full license terms →

QECTOR Decoder hero banner
Rust / Python
Stim
Sinter
PyMatching API
CUDA / OpenCL
Artifact Hashes
Belief-Matching
BP-OSD / qLDPC
Validated Evidence

Current public validation snapshot

829tests passed, 2 skipped, 1 xfailed
d=15LER parity vs PyMatching on tested workloads
33.7%lower LER at d=5 with belief-matching in the headline run
GPUCUDA/OpenCL bit-identical to CPU on tested batches
Products

Two components, one QEC platform

Available Now

QECTOR Decoder

The core QEC decoder library for researchers, benchmarking workflows, and commercial QEC evaluation. Rust core with Python bindings via PyO3.

  • Weighted MWPM / Blossom workflows with PyMatching LER parity on tested workloads
  • Belief-matching accuracy mode — 33.7% lower LER at d=5 in the headline benchmark
  • BP-OSD for LDPC / qLDPC code experiments
  • CUDA + OpenCL batch decoding, bit-identical to CPU on tested configurations
  • Stim DEM loading, Sinter plug-in, PyMatching-compatible API
  • Reproducible artifacts with SHA-256 hashes
  • 832 tests collected: 829 passed / 2 skipped / 1 xfailed
Decoder DetailsBenchmarksCommercial License
Coming Soon

QECTOR Workbench

A local fullstack app for loading circuits, running decoder comparisons, exporting artifacts, and generating reproducible benchmark reports — no cloud required.

  • Load .stim circuits and .dem files
  • Run QECTOR vs PyMatching comparisons
  • Generate PDF benchmark reports from real data
  • Export CSV / JSON artifacts with environment snapshots
  • Verify SHA-256 artifact hashes
  • CPU / CUDA / OpenCL backend diagnostics
  • Visualize LER, latency, threshold, memory, and scaling
Workbench DetailsJoin Early Access
Benchmark Evidence

Validated against PyMatching at circuit level

Rotated surface code memory_x, rounds = distance, circuit-level depolarizing noise p = 0.005, 40,000 shots per point through d=11. Same DEM decoded by both QECTOR and PyMatching.

QECTOR Blossom vs PyMatching — Stim circuit-level LER
DistanceQECTOR LERPyMatching LERQECTOR µs/shotPyMatching µs/shotResult
d = 30.01170.01170.70.5LER parity
d = 50.00890.008911.43.7LER parity
d = 70.00530.005360.413.3LER parity
d = 90.00290.0029311.519.7LER parity
d = 110.00170.0017633.946.0LER parity

PyMatching remains the latency leader for exact MWPM. QECTOR's validated advantages are reproducible evidence packaging, belief-matching accuracy mode, BP-OSD / LDPC support, GPU batch workflows, and commercial QEC integration. Read benchmark details →

Licensing

Source-available, commercially licensed

Free permitted use

Personal, academic, educational, non-commercial research

Use QECTOR for learning, private experiments, academic evaluation, benchmark reproduction, and non-commercial research under the source-available license.

Paid license required

Commercial R&D, products, SaaS, OEM, consulting, funded work

Any company, startup, institutional, government, consulting, hosted API, product integration, or revenue-linked use requires a paid commercial license.

View PricingContact admin@qector.store