Governed AI. Evidence you can inspect.
Source, running software, and model evaluations are separate claims. Follow the evidence for the exact artifact and revision; unknowns remain visible instead of being filled in.
- Explore the product — a-11-oy.com: A11oy, current capability status, and governed workflows.
- Inspect the evidence — a11oy.net: proof, receipts, evaluations, and known bounds.
- Browse the artifacts — Hugging Face / SZLHOLDINGS: models, software kernels, datasets, and Spaces. Check each artifact's evidence and limitations.
- Inspect the source estate — GitHub / szl-holdings: canonical source, evaluation lanes, and repository-specific controls.
21 public Spaces, 46 models, 35 datasets. Observed 2026-09-10T03:20:41Z under the anonymous hf-public-author-membership/v1 predicate. These are public API repository counts, not the authenticated organization total, the governed keep-list, or a count of production-ready models. Software kernels count once as model repositories; private assets, collections, and buckets are outside these totals.
Source-pinned inventory · Generation and verification · Machine-readable binding
One inference flagship: SZL Router is source-owned at szl-holdings/szl-router; its public presentation target is SZLHOLDINGS/llm-router-live, product integration is https://a-11-oy.com/code, and proof originates at a11oy.net.
Portfolio roles distinguish one inference flagship, three commercial flagships, five public domain bodies, and six internal engines. Those labels describe topology, not availability, operational readiness, or publication policy.
HISTORICAL: estate-alignment contract v1 recorded 16 portfolio Spaces, 1 inventory-only Space, 45 models, and 34 datasets. Those figures are retained only as the prior contract snapshot; they do not override the dated current public observation above. Its named topology includes A11oy, Forge, Killinchu, Terra, PRISM Counsel, PURIQ Finance, and LYTE.
Receipt verification: a11oy source contains the source and verification contracts.
Model training and evaluation: szl-forge contains kits, datasets, runbooks, measured limits, and frontier evaluation lanes.
Governance for MCP: hatun-mcp contains the governed MCP surface.
Qualification evidence: szl-frontier records exact-source frontier admissions and HOLD/EVALUATION boundaries.
Current state is a dated observation, not a promise of future availability. HISTORICAL evidence remains historical even when a newer source exists. SIMULATED results are not measured production results and must stay labeled as such.
Pre-launch. A passing source check does not establish a live deployment, and a signed record does not establish independent model quality. Runtime status, verification evidence, and remaining gaps are separate diligence inputs.
Doctrine: give away the format, sell the control plane, keep the proof.