There are 280 battery-electric variants on sale new in Australia right now, across 50 brands. The cheapest is a BYD Atto 1 at $19,999. The dearest is a Maserati GranTurismo Folgore at $450,000 — seventeen times the price. Sixteen variants sit under $40,000, which was approximately zero a few years ago.
Some things that surprised me once the data was in one place:
- The longest-range EV you can buy here is a BMW iX3 50 xDrive at 805 km WLTP. That is not a typo, and it is $109,900.
- The best value per kilometre of range is an Aion UT Premium — $74 per km of WLTP range, from a Chinese brand most Australians have never heard of. A Tesla Model 3 Long Range is genuinely excellent at 702 km for $61,900, but the value crown goes elsewhere.
- 160 of the 280 variants are SUVs. Twenty-eight are hatchbacks. Four are utes. If you want a small electric car in Australia, the market is actively working against you.
- Six brands with Australian distributors sell no battery-electric vehicle at all — Citroën, Foton, Isuzu, Jaguar (no more i-pace), Mahindra and Ram.
I built a catalogue to answer those questions, and the building turned out to be the more interesting story.
The part where there is no data
I started where anyone would: looking for an official dataset. Australia regulates vehicle emissions, runs a Green Vehicle Guide, and maintains a national vehicle register. Surely one of them publishes a machine-readable list of electric cars.
I checked all of them. Actually checked — hit the endpoints, read the terms, didn’t assume.
The Green Vehicle Guide is the only official per-model source, and it does have a live web service. It also requires a signed Third Party User Licence Agreement, obtained by emailing the Department. And even with access, it carries energy consumption and ADR test figures — not battery capacity, not charging speed, not price.
data.gov.au has no vehicle specification dataset. I queried its catalogue API directly. There are datasets for charger locations. There is nothing about cars.
The Register of Approved Vehicles is a VIN-at-a-time lookup. No bulk export, no API. Useful if you have a car and want its record; useless if you want a list.
VFACTS, the industry sales data, is behind a subscription. It was cut off even to the Electric Vehicle Council in 2024.
ev-database.org has the richest EV specification data in the world, and its terms explicitly prohibit automated collection. So I didn’t. Bulk access is a commercial licence, and scraping a site that asks you not to is not a technical problem to be solved.
That is the entire landscape. There is no open Australian dataset of electric car models. If you’ve ever wondered why every EV comparison site in this country feels slightly wrong, that’s why — everyone is reconstructing the same missing table by hand.
So I reconstructed it, and wrote down where every number came from
The only authoritative public statement about what a car sold here actually is, is the manufacturer’s own Australian website. So that became the primary source, with Australian motoring press filling documented gaps.
That approach has an obvious failure mode: it’s a pile of numbers copied off websites, and you have no way to know which ones to trust. So the central design decision was per-field provenance. Not “this car came from Kia” — every individual figure carries its own publisher, URL, fetch date and confidence level.
Open any vehicle in the explorer and click Sources, and you get a row per field. You can see that the Kia EV5’s battery capacity came from Kia’s own brochure PDF, that its DC charging rate came from CarExpert and is flagged lower-confidence, and that its price is a drive-away figure rather than MSRP. You don’t have to trust the catalogue. You can check it.
Where two sources disagreed by more than 2%, the more authoritative one wins and the rejected value is recorded rather than deleted. Five such conflicts exist in the current build. Honda’s Super-ONE is the nicest one: Honda’s own page quotes 253 km on the NEDC cycle, while its later pricing announcement quotes 206 km WLTP. Those are different tests, not a correction, so both are kept and neither is averaged into a comforting middle number.
Three rules that made it honest
A figure nobody publishes is null. Never zero, never an estimate, never borrowed from
the European version of the same car. It renders as an em dash. 183 of 280 variants have
unknown vehicle-to-load capability, because most brands simply don’t say.
Range figures are never merged across test cycles. WLTP, NEDC and CLTC produce different numbers for the same car, and a Chinese-market CLTC figure flattered into a WLTP column is a lie with a decimal point. The cycle is stored and displayed alongside the number.
Numeric filters exclude unknowns. Filter to “under $50,000” and you will not silently get a car whose price is unpublished — because a filtered list that includes unknowns implies specs the catalogue doesn’t have.
Completeness is a measurement, not a boast
I can’t claim the catalogue is complete, so instead it reports how complete it is: 89.9% of core fields populated. The per-field breakdown ranges from 100% (brand, model, body type, availability) down to 22.8% for published energy consumption — barely any manufacturer states it.
The provenance mix is worth stating plainly too: 79% of populated fields come from manufacturers, 21% from motoring press, and 0% from any government source. That last number is the whole story of this project in one figure.
And the six brands that sell no EV get shipped as files with empty vehicle lists, because “checked, sells nothing” and “never looked” are different claims, and a catalogue that can’t tell them apart isn’t trustworthy.
What it caught in itself
I wrote a validator with ESLint-style severities over the data. It found three real defects, and I like that it found them more than I like any feature:
-
A 12-seat Skywell van failed the seat-count check. The van was fine. My plausibility limit of 9 seats was wrong, because I’d been thinking about cars and the catalogue includes commercial vans.
-
A Denza Z9GT tripped the charging-speed ceiling at 1,500 kW. I assumed a transcription error. It isn’t — BYD genuinely publishes that figure for its FLASH Charging system. But it’s only reachable on a China-market connector, and no Australian charger comes close, so the record now carries that caveat instead of implying you can get it here.
-
The filter panel was impossible to open on a phone. The button had
display: nonefrom a base CSS rule sitting after the media query meant to reveal it, so it lost the cascade at every screen size. Every test passed. It took actually loading the page at 390 pixels wide to see it.
The last one is the lesson. Ninety-four passing tests told me the data layer was sound and said nothing whatsoever about whether a human could use the thing.
Using it
The explorer is a single self-contained HTML file — search, filter, sort, compare four cars side by side, inspect any figure’s source. No server, no build step, works offline. There’s also a command-line tool and a JSON bundle for anyone who wants the data rather than the interface.
Two caveats I’d rather say than bury. Prices are indicative and exclude on-road costs unless the record says otherwise; 39 records carry drive-away figures because that’s all the brand publishes, and drive-away varies by state. And the code is open while the data is not — specification figures are restated facts attributed to their publisher, not open-licensed data.
Keeping it current means re-checking manufacturer websites, because there is no feed to poll. That’s the cost of a market with no open dataset, and it doesn’t go away until someone publishes one.
FAQ
How many electric cars are sold in Australia?
280 battery-electric variants across 50 brands, as of September 2026.
What’s the cheapest EV you can buy in Australia?
A BYD Atto 1, at $19,999.
What’s the longest-range EV available in Australia?
A BMW iX3 50 xDrive, rated at 805 km WLTP, priced at $109,900.
Which car brands don’t sell any EV in Australia?
Six brands with Australian distributors sell no battery-electric vehicle: Citroën, Foton,
Isuzu, Jaguar, Mahindra and Ram.
Where does the data in the catalogue come from?
Mostly manufacturers’ own Australian websites (79% of populated fields), with Australian
motoring press filling documented gaps (21%). No figure comes from a government source.
Every field carries its own publisher, URL, fetch date and confidence level.
Is the catalogue’s data free to reuse?
The explorer’s code is open, but the specification data is not — it’s restated facts
attributed to their original publisher, not open-licensed data.
How complete is the catalogue?
89.9% of core fields are populated overall. Coverage varies a lot by field: 100% for
brand, model, body type and availability, down to 22.8% for published energy consumption.
