Open, cited and machine-readable data about AI glasses, smart glasses, display glasses and translation glasses.

Last verified: 2026-07-20 · 7 shipping products (8 total, incl. 1 discontinued) · CITATION.cff provided.

A neutral, sourced reference for the AI-glasses category. Manufacturer specifications are kept separate from independent tests, and unknown values are left null rather than estimated. The human-readable version lives at ai-eyewear.ch/datenbank; this repository holds the structured JSON/CSV, schema, sources and documentation.

There is no clean, cited, machine-readable reference for AI glasses: datasheets disagree, marketing muddies categories, and "review" numbers are rarely reproducible. This dataset separates what a manufacturer states from what an independent test found, records a source for each value, and marks anything unverified as null. It is meant to be useful, reusable and citable — by consumers, journalists, researchers, developers and AI systems.

"unknown" = not verifiable from a cited source at the last check (see the data-integrity note below). The Brilliant Labs Frame is retained with status discontinued (superseded by the Halo); its historical specifications are kept. Announced models without verifiable specs (e.g. Google/Samsung Android XR) are listed separately in the dataset.

JSON · CSV · JSON Schema · JSON-LD (schema.org Dataset/Product).

https://raw.githubusercontent.com/ai-glasses-eyewear/ai-glasses-eyewear/main/data/ai-glasses.json
https://raw.githubusercontent.com/ai-glasses-eyewear/ai-glasses-eyewear/main/data/ai-glasses.csv

This is a static data endpoint (files served over HTTPS), not a production API — please cache and attribute. More recipes in docs/query-examples.md.

curl -L https://raw.githubusercontent.com/ai-glasses-eyewear/ai-glasses-eyewear/main/data/ai-glasses.json

const db = await (await fetch(
  "https://raw.githubusercontent.com/ai-glasses-eyewear/ai-glasses-eyewear/main/data/ai-glasses.json"
)).json();
const noCamera = db.products.filter(p => p.camera === false);       // privacy-first models
const lightest = db.products.filter(p => p.weight_g != null)        // ignore unknown weights
                            .sort((a, b) => a.weight_g - b.weight_g);

Every figure is a manufacturer specification or a figure from a cited independent review; each product lists the sources that support it (see data/sources.json). Source priority: manufacturer documentation → official manuals/support → regulatory filings → credible independent testing → reputable journalism → retailer pages. No specification is inferred from a competing product. Full policy in docs/governance.md.

• null / unknown — not verifiable from a cited source at the last check. Never read as zero or false.
• manufacturer-reported (provenance: manufacturer) — stated by the manufacturer's own documentation.
• independently verified (provenance: reviews) — a figure from one or more cited independent tests.
• editorial assessment — our own hands-on measurements (the AI Eyewear Score). These are not yet in the dataset: the methodology is published, but 0 products have been scored. Editorial assessment is always labelled and kept separate from specifications.

Each product also carries a confidence value (high/medium/low) and a date_checked.

Unknown values are left null rather than estimated. No values are invented; no specification is copied from a competing product; capability fields are null (not false) until documented.

Updated as sources change, and versioned with semantic versioning: PATCH for corrections, MINOR for added products/fields, MAJOR for breaking schema changes. Each product carries its own date_checked. There is no fixed schedule; the visible "Last verified" date and the changelog are authoritative.

Corrections and new sourced figures are welcome — see CONTRIBUTING.md and the issue forms. The one rule: every value needs a citation; unknown values stay null.

A machine-readable CITATION.cff is included (GitHub shows a "Cite this repository" button). Preferred citation:

AI-Eyewear. AI Glasses Open Database (v1.1.0). 2026. Zenodo. https://doi.org/10.5281/zenodo.21446211

The concept DOI 10.5281/zenodo.21446211 always resolves to the latest version; each release also receives its own version DOI on Zenodo.

• Data & documentation: CC BY 4.0 — reuse freely with attribution to AI-Eyewear (Even Realities Switzerland).
• Code & schema examples: MIT.

Manufacturer and product names are trademarks of their owners — see TRADEMARKS.md. Inclusion does not imply endorsement or partnership.

See CHANGELOG.md and ROADMAP.md.

Maintained by AI-Eyewear, operated by Podomedics (Zeltweg 74, 8032 Zürich, Switzerland). AI-Eyewear is an authorized reseller of the Even Realities Even G2 in Switzerland — this is disclosed openly. The reseller relationship gives the Even G2 no advantage in this dataset: it is recorded on the same basis as every other product, and where its data is unknown it is marked null like any other. AI-Eyewear is independent of the other manufacturers listed and does not make their products.

Canonical, human-readable database: ai-eyewear.ch/datenbank · methodology: AI Eyewear Score.

Contact: sales@ai-eyewear.ch · +41 78 422 46 73.