https://benthic.io/parts/NHTSA/ – NHTSA Vehicle Product Information Catalog

The NHTSA dataset combines the official vPIC decoder data with the vPIC reference data used to interpret vehicles, manufacturers, makes, models, WMI codes, variable values, and equipment plants. The decoder and reference data are published through benthic.io; reference tables are synchronized from the vPIC API.

The default profile contains the decoder tables and RPCs. Reference tables are published at the /reference/ path.

At a glance

Quick start

# Decode one VIN
curl -X POST "https://benthic.io/parts/NHTSA/rpc/spvindecode" \
  -H "Content-Type: application/json" \
  -d '{"v":"1HGCM82633A004352"}'

# Decode a batch
curl -X POST "https://benthic.io/parts/NHTSA/rpc/spvindecodemultiple" \
  -H "Content-Type: application/json" \
  -d '{"vin_list":["1HGCM82633A004352","1FTFW1ET4EFA12345"]}'

# Query the upstream decoder tables
curl "https://benthic.io/parts/NHTSA/wmi?limit=1"
curl "https://benthic.io/parts/NHTSA/make?select=id,name&order=name&limit=20"

# Query synchronized reference data
curl "https://benthic.io/parts/NHTSA/reference/manufacturers?limit=25"
curl "https://benthic.io/parts/NHTSA/reference/wmi_codes?select=wmi,brand_name,country&limit=25"

The decoder returns one row per decoded variable. The Error Code and Error Text variables report whether the VIN decoded cleanly. A VIN that is syntactically valid but not present in the current vPIC patterns may return a non-clean result.

Data profiles

The published reference path selects the api_reference profile automatically. Direct PostgREST clients can select it with Accept-Profile: api_reference on GET requests and Content-Profile: api_reference on writes.

Tables

Decoder tables

Use the OpenAPI specification for the complete relation and column contract.

Reference tables

Querying reference data

# Search manufacturers
curl "https://benthic.io/parts/NHTSA/reference/manufacturers?manufacturer_name=ilike.*HONDA*&select=manufacturer_id,manufacturer_name,manufacturer_common_name&limit=25"

# Find WMI codes for a manufacturer
curl "https://benthic.io/parts/NHTSA/reference/wmi_codes?manufacturer_id=eq.2364&select=wmi,brand_name,country&order=wmi&limit=100"

# List active equipment plants in a country
curl "https://benthic.io/parts/NHTSA/reference/equipment_plants?plant_country=ilike.*UNITED STATES*&plant_status=eq.Active&select=plant_name,plant_city,plant_state,dot_code&limit=50"

# Historical models for a make
curl "https://benthic.io/parts/NHTSA/reference/models_historical?make_id=eq.468&select=year,model_id,model_name&order=year.desc&limit=100"

PostgREST query reference

The same syntax as every other benthic.io dataset. Paths differ: decoder tables and RPCs are at the root, reference tables under /reference/.

Filtering

Case-insensitive matching: use ilike, not like, when the pattern casing is unknown. ilike treats * as the wildcard; the pattern must be URL-encoded, so * is %25 in a query string.

Selecting columns

?select=manufacturer_id,manufacturer_name

Omitting select returns every column, which on models_historical is substantially more data than you usually want.

Ordering

?order=year.desc
?order=manufacturer_name.asc,year.desc

Pagination

?limit=100&offset=200

Or use range headers:

Range: 0-99

Counting

Prefer: count=exact

?select=count

Grouping / aggregation

?select=plant_state,count&groupby=plant_state&order=count.desc

Calling the decoder functions

The two vPIC decoder RPCs take a JSON body and are called with POST:

curl -X POST "https://benthic.io/parts/NHTSA/rpc/spvindecode" \
  -H "Content-Type: application/json" \
  -d '{"VIN": "1HGCM82633A004352"}'

curl -X POST "https://benthic.io/parts/NHTSA/rpc/spvindecodemultiple" \
  -H "Content-Type: application/json" \
  -d '{"VINs": ["1HGCM82633A004352", "5YJ3E1EA8LF000000"]}'

Provenance and limits

• The decoder tables and functions are published through benthic.io from the official NHTSA/vPIC dataset.
• Reference tables are synchronized from the vPIC API and published under /reference/.
• The published decoder data is a snapshot; some current vPIC API responses may differ.
• Plant records without a stable DOT code are omitted from the reference tables.
• NHTSA/vPIC data is a U.S. government work; check the NHTSA site for applicable terms and attribution requirements.

The signed BDP manifest and collection describe the published schema and provenance.

Key relationships

NHTSA sits in the parts collection and shares no key with the NGOpen datasets — there is no join from vehicle reference data to federal spending, nonprofit, or legislative data. What it does offer is a lookup direction:

• spvindecode is a bridge from an arbitrary string to structured data. A VIN that appears in a fleet operator’s records, an insurance claim, or a recall notice can be decoded to make, model, year, and manufacturer without knowing any of them in advance. That is the intended entry point.
• Manufacturer names are the loose join to everything else. manufacturers and wmi_codes carry name and country columns, so a manufacturer observed in samer or usaspending can be matched by name — but this is a fuzzy entity-resolution problem, not a key. Do not treat it as one.
• The WMI prefix encodes the manufacturer. The first three characters of a VIN are the World Manufacturer Identifier and appear in wmi_codes, so a partially-redacted VIN can often be resolved to a manufacturer without a full decode.

Data sources

Decoder data is a snapshot of the current vPIC release and is refreshed monthly. Reference tables are synchronised from the API and published under /reference/; some current vPIC API responses may differ from the published snapshot.

All data originates from publicly available U.S. government sources. NHTSA/vPIC is a U.S. government work — see the NHTSA site for applicable terms and attribution requirements.

Related

• Swagger explorer — interactive API browser for /parts/NHTSA
• Signed manifest — every relation, column, and licence, signed with the publisher’s Ed25519 key
• Pipeline README — how this dataset is actually built
• BDP specification — what the manifest signature proves, and how to verify it offline
• Documentation map — every dataset on benthic.io and where its documentation lives
• APIs overview — join paths between datasets and worked cross-collection queries
• In Ngopen: USAspending, SAM Entity Registry, IRS Nonprofits, Congress Legislators, Congressional District Maps