Coverage ledger — every.autos

No public source defines every car ever built, and “model” can mean a nameplate, generation, trim, market alias, prototype, or model year. We will not turn that ambiguity into a false 100% badge. Completeness here means reconciling a named, versioned source universe and showing the evidence behind every field.

The content percentages and identity-contract counts below are deliberately separate. A record can have useful history and specifications while its semantic level, category, alias review, or scope decision still needs work.

Catalog records: 13,147
Marques represented: 1,599
Complete across all 7 content dimensions: 5,180
Missing at least 1 content dimension: 7,967

Current snapshot

What the records actually contain

Sourced historyCatalog description, validated Wikipedia background, or reviewed editorial

81.1%10,656 / 13,147
Production chronologyA sourced production start; market observations do not qualify

74.2%9,761 / 13,147
OriginModel or marque origin, qualified and kept separate from registration market

98.6%12,965 / 13,147
Vehicle classA sourced class or body-style assertion

96.0%12,616 / 13,147
Technical factsSpecs, powertrain, production, economy, or other scoped evidence

65.3%8,584 / 13,147
ImageryA record image from Wikimedia Commons or another named source

72.7%9,563 / 13,147
Multiple datasetsThe identity appears in more than one ingested source

57.9%7,612 / 13,147

Inclusion contract

What “complete” will mean

The working catalog centers on automobiles and qualifying complete road vehicles, organized as marque → nameplate → generation → market variant. The current 8-item unresolved scope-review queue flags legacy admitted records whose vehicle-class or body-style classifications contain selected two-wheeler or powersports terms, such as motorcycle, moped, scooter or autocycle. Each remaining flag is a review candidate, not an automatic exclusion. The pinned ledger has already resolved 283 other classifications: 22 included and 261 excluded. Trailer, caravan, component and other non-road records are not measured by this queue.

4,162 identity-field records still need review 8 legacy classifications remain in the two-wheeler or powersports review queue 13,762 records still need explicit nameplate, generation, or variant placement

01
Stable identity fields
9,608 / 13,770 pipeline records satisfy the currently measured alias, source-entity-level, category, and provenance fields (69.8%). The pipeline ledger is broader than the public global projection.
02
Qualified chronology
Production, model year, sale date, and registry observation stay distinct.
03
Geography with meaning
Origin, assembly country, sales market, and registry market are separate facts.
04
Field-level provenance
Every accepted assertion retains its source, retrieval date, licence, and confidence.
05
Alias review
9,608 identities reviewed; 1,683 currently retain one or more source aliases.
06
Useful depth
A sourced history plus technical, production, or market facts—not generated filler.

Expansion map

Measured universes and evidence feeds

Licensing & methods →

AllCarIndex brand identities14,977 public identities · 44 reviewed admissions · 1 reasoned classifications · 14,932 awaiting review, including 356 independently present in Wikidata Wikidata manufacturers6,917 global candidates · 1,387 stable catalog links, including 436 explicit admissions · 189 reasoned exclusions · 5,341 awaiting scope review Natural Resources Canada universe5,260 candidate make/model identities · 669 exact matches · 4,591 family, prefix, or contained-name rule matches · 0 awaiting review under current rules NHTSA vPICUS manufacturer and model identity; current admissions use catalog heuristics while an exact category-partition audit remains backlog EPA FuelEconomy.govUS model-year, powertrain, class, economy and emissions assertions EEA CO₂ monitoringEU commercial names, type/variant/version and approval data since 2010 Netherlands RDWRegistered trade names, dates, approvals and linked technical facts Swiss TARGAHistorical type approvals extending toward 1946

Each dataset measures a market and period—not the whole world. Every upstream row remains accounted for as a current outcome or a pending review; the universe is not fully resolved while any review queue remains. Automated triage labels are proposals, not admission or exclusion decisions.