Start free. Upgrade when the decisions start mattering.

Qlro is in the free-adoption phase. Our moat is the paired (prediction, hardware outcome) dataset that accumulates as users contribute back observations. We want as many quantum engineers as possible to run Qlro on their workloads before we turn on billing.

Existing free-tier keys stay free forever at the 500 req/month quota. If you joined the Team waitlist before Q3 2026 you lock in the $49/seat pricing with a 20% permanent discount (so 2 seats stay at $78.40/mo for the lifetime of the subscription). Enterprise contracts are negotiated individually.

Quantum-device decisions are organisational, not personal — we haven't seen demand for a per-seat individual-professional tier in this market. Free covers solo research; Team starts at 2 seats ($98/mo) for small pilot groups. If enough individual demand shows up we'll revisit.

Yes, via Enterprise. The full predictor + dashboard runs in a private container with your own Metriq snapshot and a private audit-trail database. Contact sales.

Enterprise contracts start at $50K/yr (base) and stack with five add-on lines: Audit + Reporting (+$30K), Private Deployment (+$50K), Error Mitigation Advisory (+$80K), Procurement Advisory (+$100K), and Continuous Drift Monitoring(+$120K). A government-pilot contract typically lands at $230K ACV; an R&D fleet contract typically at $250K ACV. See the Enterprise page for the add-on detail and the 3-year vision roadmap.

The SDK ( github.com/linsletoh/qlro, Apache 2.0) and the paper (Zenodo DOI 10.5281/zenodo.19785800) are fully public. The hosted API / dashboard / pricing tiers are commercial.

Nothing — and we don't want to stop them. The paper and the four-axis projection are the ingredients; the moat is the paired (prediction, real-hardware outcome) dataset that compounds with every user submission, plus the citation graph that forms when external papers, RFPs, and procurement records reference Qlro snapshot DOIs. A cloned formula with no residual time-series and no cited snapshots produces numbers nobody can reproduce or audit.

Because a vendor-run comparison is structurally non-credible. The value of a cross-vendor recommendation comes from the neutrality of the party producing it — the instant IBM ranks IBM against Quantinuum, buyers discount the result. This is why Bloomberg could build a cross-market terminal that JPM and Goldman couldn't: the neutral party wins the reference-point role.

Metriq is the benchmark layer; Qlro is the workload-conditioned decision layer on top of it. The two compose rather than compete — we cite Metriq in every snapshot. If they ship a ranker, we ingest it as a signal; if they don't, we keep scoring their data.

We publish the residuals. Every prediction is timestamped against the Metriq commit it was made on, and the corresponding hardware outcome — when users contribute one back — is diffed publicly on /accuracy. A predictor that hides its misses has no feedback loop; one that shows them compounds calibration over time.