Each dot is one product. Shape encodes technology class — electric motor, combustion engine, turbine, propeller — so two finely-different technologies in the same class share a glyph and the legend labels are finer than the shapes (see the "i" button next to the Technology selector for definitions; 2026-08-29: shape had silently fallen back to a circle for every row corpus-wide, and this sentence said "technology" while the encoding was already class-level). Color encodes sourcing region. Every dot carries a thin dark outline (keeps overlapping points distinguishable); outline weight encodes how many independent sources verify the row: thin = 0, hairline = 1, medium = 2, heavy = 3+.

• datasheet — vendor datasheet URL was directly verified at scrape time.
• FCC brand — manufacturer holds at least one FCC grantee code (US-marketed RF gear).
• FCC equipment — a specific FCC ID was matched, with grant-level frequency band & output power.
• FAA DOC — manufacturer has an accepted FAA UAS Declaration of Compliance (Part 89 Remote ID or Part 107 OOP).

Confidence tier in the detail panel (same field as the open dataset) grades the performance claim only — country and price carry their own evidence grades, shown beside it (a row can hold a Tier-0 performance figure while its country claim is unevidenced): Tier 0 Anchor = performance figure anchored to a held manufacturer datasheet or independent bench test; Tier 1 Verified = corroborated by two or more independent source domains; Tier 2 Single-source = vendor/distributor listing, one source; Tier 3 Inferred = inferred, or no measured performance figure. Use the Evidence-tier filter above to restrict to any tier.

Power basis. Vendor wattages mix electrical input, electrical output and mechanical shaft figures, which differ by 1.23–1.77× depending on throttle and cannot be converted. Derived axes (power-to-weight, thrust estimates) are therefore computed within one stated basis; rows whose basis the vendor did not state are excluded and counted in an on-chart note, and each point's basis is shown in its tooltip and detail panel. The same discipline applies to power duration class: a burst or peak figure is not comparable with a continuous rating, so the derived power-to-weight axes admit only continuous/180s/rated figures (untagged and burst-class rows are excluded and counted), and the plain Max-power axis shows each vendor's largest published figure with its duration class — where established — in the detail panel.

Caveat: only 22 of the 376 rows in this explorer have ≥ 2 independent sources. Treat single-source rows as preliminary, and weight aggregate/share statistics by tier and source health.