Car Specs API — 103k+ vehicle variants, 20 languages, EV data

The car-specs data source built for AI agents. Search, full localized specs, side-by-side compare, and structured filters over 103,372 vehicle variants, 1,357 models, 5,428 generations, 230 spec types, across 20 languages (including Arabic, RTL) — from cars-data.com.

No other car-specs provider covers this many European generations in this many languages. This Actor is also auto-exposed as an MCP tool, so it works directly inside Claude, Cursor, and Windsurf — not just the Apify Store.

At a glance

• 4 modes — search, get_specs, compare, filter, one Actor
• 20 languages — including Arabic (RTL)
• EV & hybrid ready — dedicated fuel/power filters, battery + range specs
• 230 spec types per variant, each with a confidence score and last_synced_at

What it does

Pick a mode and this Actor calls the matching cars-data.com endpoint and pushes the result to the run's dataset:

mode	What it does	Required input
search	Free-text search by brand/model/generation name	query
get_specs	Full localized specs for one variant — 230 spec types, with confidence + freshness	variantId
compare	Side-by-side specs for 2-4 variants	variantIds
filter	Structured filter: fuel, body, drive, power, price, year, EV	(all optional)

Every get_specs/compare result carries a confidence score per spec value and a last_synced_at timestamp — so you can be honest about data freshness instead of presenting stale or uncertain values as fact.

What's in the 230 specs

Category	Specs	Examples
Safety	32	ABS, airbags, ADAS, crash rating
Interior	32	seat material, layout, storage
Fuel & Engine	24	displacement, cylinders, injection
Exterior	22	lights, wheels, roof, body trim
Comfort	15	climate control, seats, infotainment
Exterior Sizes	15	length, width, height, wheelbase
Chassis	14	suspension, brakes, steering
Electric / Hybrid	14	battery capacity, range, charging, heat pump, V2L/V2H/V2G
Performance	14	0-100, top speed, power-to-weight, derived ratios
Interior Sizes	12	headroom, legroom, shoulder room
Weights	9	kerb weight, payload, towing
General	6	body type, doors, seats
Consumption (WLTP)	4	combined, low/medium/high phases
Consumption (EPA)	3	city, highway, combined (US cycle)
Consumption (NEDC)	3	combined/city/highway, older cycle
Luggage / Load	3	boot capacity, sill height
Service & Warranty	3	service interval, warranty terms
Costs & Taxes	1	road tax band
Drive	1	driven wheels
Emissions	1	CO₂
Transmission	2	gearbox type, number of gears

All 21 categories, summing to exactly 230 — the number of spec types that actually carry data in production (232 are defined; two carry no value yet). Counted on cars_prod, not estimated.

Example input — search

{
  "mode": "search",
  "query": "bmw 3 series",
  "locale": "ro",
  "limit": 5
}

Example output (one item, search for "tesla model 3"):

{
  "variant_id": 108748,
  "display_name": "Tesla Model 3 Performance AWD",
  "brand_slug": "tesla",
  "model_slug": "model-3",
  "generation_id": 7608,
  "year_from": 2023,
  "year_to": 0
}

Example input — full specs

{
  "mode": "get_specs",
  "variantId": 42164,
  "locale": "de"
}

Example output (specs trimmed — a real run returns every spec field the variant carries, up to 230, each with value, unit, and confidence; the other fields are the actual headline numbers, not buried in specs):

{
  "variant_id": 108748,
  "locale": "en",
  "display_name": "Tesla Model 3 Performance AWD",
  "brand_slug": "tesla",
  "model_slug": "model-3",
  "year_from": 2023,
  "year_to": 0,
  "power_hp": 627,
  "torque_nm": 741,
  "top_speed_kmh": 262,
  "accel_0_100_s": 3.1,
  "battery_kwh": 75,
  "price_new_eur": 57490,
  "specs": {
    "abs": { "label": "ABS", "value": "Standard", "unit": null, "confidence": 0.9 },
    "width_mm": { "label": "Width", "value": 1849, "unit": "mm", "confidence": null },
    "navigation": { "label": "Navigation", "value": "Standard", "unit": null, "confidence": 0.9 },
    "payload_kg": { "label": "Payload", "value": 442, "unit": "kg", "confidence": null }
    // ...and the rest of this variant's spec keys
  }
}

That same shape also drives the Overview table view — get_specs runs render brand, model, power, torque, top speed, battery and price as columns, not just raw JSON.

Example input — compare

{
  "mode": "compare",
  "variantIds": [42164, 42165],
  "locale": "en"
}

Example output: an array of the same shape shown above, one entry per compared variant — so you can diff headline numbers or spec-by-spec across 2-4 cars.

Example input — filter (electric, 200+ hp)

{
  "mode": "filter",
  "fuel": "electric",
  "powerMin": 200,
  "limit": 10
}

Example output (real EV-filter results):

[
  {
    "variant_id": 60418,
    "display_name": "Tesla Roadster",
    "power_hp": 252,
    "battery_kwh": 53,
    "top_speed_kmh": 200,
    "accel_0_100_s": 3.9,
    "fuel_slug": "electric",
    "body_type_en": "Convertible",
    "price_new_eur": 99960
  },
  {
    "variant_id": 60419,
    "display_name": "Tesla Roadster Sport",
    "power_hp": 292,
    "battery_kwh": 53,
    "top_speed_kmh": 200,
    "accel_0_100_s": 3.7,
    "fuel_slug": "electric",
    "body_type_en": "Convertible",
    "price_new_eur": 117810
  }
]

That same data, as the Actor's built-in Overview dataset view renders it:

Pricing

Pay-per-event — no subscription, no rental. You are charged only for completed runs, per the event triggered by your chosen mode (search, spec-lookup, compare, or filter). See this Actor's Pricing tab for exact per-event rates.

Data & attribution

Data comes from cars-data.com's multilingual vehicle-specs database. Output is licensed for use within your own application/workflow, not for resale or reconstruction of the underlying dataset — see cars-data.com/api/terms.

Try it first: a free MCP server, 40 cars

There is also a remote (streamable HTTP) MCP server at https://api.cars-data.com/mcp — no local install, no API key, connect it directly from your AI assistant. It exposes the same 6 tools: search_cars, get_specs, compare_variants, filter_cars, list_generations, get_images.

It answers from a fixed 40-car sample, not from the catalogue. Five brands x two models x two generations x two variants, covering all 10 fuel types, with every spec, every image and all 20 languages — enough to see exactly what the data looks like before you pay for any of it.

This Actor is the full catalogue — 103,372 variants, billed per result. If you need more than the sample, that is what you are here for. For a one-off bulk export rather than per-lookup reads, see cars-data.com/en/api.

Cursor — add to .cursor/mcp.json (Cursor Settings → MCP → Add):

{
  "mcpServers": {
    "cars-data": {
      "url": "https://api.cars-data.com/mcp"
    }
  }
}

Windsurf — same shape, in ~/.codeium/windsurf/mcp_config.json (Windsurf Settings → Manage MCPs → View raw config):

{
  "mcpServers": {
    "cars-data": {
      "url": "https://api.cars-data.com/mcp"
    }
  }
}

Claude (claude.ai or Claude Desktop) — Settings → Connectors → Add custom connector → paste https://api.cars-data.com/mcp.

ChatGPT — Settings → Connectors → Advanced → paste https://api.cars-data.com/mcp as a custom MCP connector (availability depends on your ChatGPT plan/workspace).