Vehicle record · Australia

BYD ATTO 3 EVO Premium.
Three admitted measurements from one dated capture.

One trim of one generation, recorded from a single capture of the manufacturer’s Australian model page: battery capacity, maximum power, and the driving-range row the source labels NEDC. Those three figures are admitted here. Everything else the same capture displays — the other trim columns, the prose range sentence on a different test procedure — is either scoped away or explicitly held.

Vehicle BYD ATTO 3 EVO
Trim Premium
Market Australia
Captured 22 Sep 2026
Status Not live — snapshot

01 / Admitted measurements

Three figures, one trim, one snapshot. Each one is bound to the panel row it came from; nothing on this page is filled in.

Battery capacity 74.88 kWh Source row label: Battery capacity. Admitted for the Premium trim of the EVO generation only, as printed in the captured panel.
Maximum power 230 kW Source row label: Maximum power. The word maximum belongs to the label and is kept: this is a peak output claim, not a continuous rating.
Driving range, source label: Driving range NEDC 600 km Source row label: Driving range NEDC. Read the figure inside the NEDC procedure only; the source does not state the conditions or the mode it was obtained in.
Driving range under the WLTP procedure Under review — not admitted The same capture carries a prose sentence about a WLTP figure for this trim. It is not a panel literal, its wording does not fix an exact distance, and it names a different procedure than the admitted row. It stays held, and nothing from it is published here.

Three admitted measurements are not a specification sheet. The capture shows more than this page does, and the difference is deliberate: an attribute the pinned admission policy does not allow stays unrendered rather than being paraphrased into a softer claim.

02 / What the NEDC label supports

The one test procedure this record can name, and the limits of naming it.

A laboratory cycle, not a road test

The range row on the cited page reads “Driving range NEDC”. That label is the only test procedure the source attaches to the figure, so it is the only one this record can report. NEDC names a laboratory driving cycle: a defined sequence of accelerations, steady speeds and stops that a vehicle performs under controlled conditions. A distance produced inside that sequence describes behaviour inside that sequence. It is a regulatory artefact with a fixed shape, and reading it as a description of ordinary driving inverts what it is.

What it cannot be asked to prove

A NEDC-labelled distance does not establish how far this vehicle travels on a highway, in dense traffic, in summer heat, with passengers, with a load, or with climate control running. It does not describe a worst case or a typical case. This page therefore makes no statement about everyday travel, and it does not convert the figure into another procedure’s units or into an expectation dressed up as a converted result. Turning one cycle’s number into another cycle’s number would be an invention with a decimal point on it.

Scope wording is quoted, not upgraded

The row label also carries the words “Driving range”, and those words are preserved as scope wording rather than being read as a combined, electric-only or total figure. The source does not say which conditions, which mode or which measurement boundary the distance belongs to, so the record leaves that unstated instead of choosing the most flattering reading available. Where a label is narrow, the honest move is to keep it narrow and let a reader see the narrowness.

No cycle is ranked against another

Nothing here orders the admitted figures, and no value from one procedure is placed against a value from another as though the two were the same measurement twice. Ranking requires a common basis; different driving cycles are not a common basis. Since this page carries one procedure-labelled range row and one held observation, it has no ordering surface at all — not a sort, not a badge, not a superlative about range, battery or output.

03 / Trim and generation scope

Why the trim sits in the page path instead of hiding in the prose.

The path names the trim on purpose

This record speaks for one entity: the Premium grade of the EVO generation of this model, in Australia. The captured panel shows several columns, and the admitted figures come from the column whose card identity carries the model line and the trim name together. Publishing those numbers as model-wide facts would quietly promote one grade’s figures into a claim about every version of the vehicle, which is exactly the promotion this page refuses. A figure belongs to the configuration it was measured on, and the address of this page says so.

Other columns are visible but unadmitted

The same panel lists a second grade of the EVO generation and a column whose header does not name that generation at all. Neither is admitted, and neither is rendered with figures here. The second is excluded because this admission is bounded to a single trim; the third is excluded because a shared trim name across generations is no identity proof, and a column header that omits the generation cannot be used to attribute its literals to the EVO vehicle. Being visible in a capture is not the same as being admitted from it.

A footer notice is not a generation proof

The cited capture also carries a generic notice describing its specifications as belonging to the model line. That notice names a family, not a generation, so it is not treated here as evidence about which vehicle the figures describe. The generation identity in this record rests on the comparison card the admission is bound to; the generic notice is recorded as what it is and nothing is inferred from it. Where the source is ambiguous, the page reports the ambiguity instead of resolving it by preference.

Peak output stays labelled as peak

The power row appears under the label “Maximum power”, and the qualifier attached to the admitted measurement records that the figure is a maximum. That distinction matters because a peak figure and a continuous rating describe different physical properties and are not interchangeable. Keeping the label’s own word attached to the value is the cheapest way to stop a peak claim from being read as a sustained one.

04 / Snapshot limits and held material

What a dated capture can carry, and what it cannot.

One capture, one date

Every figure above was what the cited source displayed on the capture date, and the source may have changed since. Nothing on this page is fetched or refreshed when it is read, no brochure, price schedule or later revision has been merged into the record, and no model year is claimed. The capture date and the snapshot hash sit in section 05 so any figure can be checked against the bytes it came from rather than taken on trust. Where verification was not possible, the record says so instead of smoothing the corners of the unknown.

The held range observation

The same capture contains a prose sentence about a range figure for this trim measured under a different procedure. That sentence is not an exact panel literal, its own wording does not fix a distance, and it belongs to a cycle the admitted row does not use. It is therefore held: not admitted, not rendered as a cell, not averaged with the admitted distance, and its numeric literal is withheld from this page and from the machine-readable record. The cell in section 01 reads “Under review” so that the gap is visible rather than filled with something plausible.

No pricing conclusion of any kind

This page draws no pricing conclusion and states no commercial figure for this vehicle or any other, in Australia or elsewhere. The admission policy behind these measurements excludes that category by design: an initial price and any change to a price require first-stage human approval, and a deterministic policy may not admit, bind or project one. That exclusion is why the record carries no amount and no currency rather than carrying a number it cannot ground. When a human-gated commercial admission exists, it will arrive through its own reviewed surface — not through a line quietly added here.

What is not measured here

Attributes the capture displays but the pinned allowlist omits — drivetrain wording, acceleration, kerb weight, luggage capacity, charging prose and warranty terms — stay unadmitted and unrendered. Their absence is not a judgement about them; it is a statement about the evidence that has passed admission so far. A short record whose every line can be traced is more useful than a long one whose lines cannot, and this page is built to stay on the short side of that trade.

Published unindexed, on purpose

This record ships marked noindex with a self-referencing canonical address while it is a research preview. The content is complete and its status is stated rather than hidden: the page is published for people who arrive at it, and its indexing decision belongs to a separate review that has not been taken. A preview that says so openly is easier to audit than one that pretends to be finished.

05 / Behind the numbers

Trace each figure to its capture, its panel row and its snapshot hash.

BYD ATTO 3 EVO Premium — Australia · capture of 22 Sep 2026
Capture date
Source label
BYD Australia — model page (captured final address: bydautomotive.com.au)
Snapshot SHA-256
158b9a2238f012d31cb1981289085dd0838a0150039221afe392a7b66e9b3a18
Locator
HTML comparison-card alt="BYD ATTO 3 EVO Premium" > ul.comparison-bottom > span.title="Battery capacity"
Admission artifact
ChinaEV-Truth-Plane-Spike/metadata/ATTO3_EVO_ADMISSION.json (sha256 95104d643136d826f3b5c7a05afec70211c6880935f47e44d299f3edd71cd54c, status ADMITTED)
Structured data
evidence.json (the pinned evidence projection of this record: one trim-scoped record, three admitted measurements)

The snapshot itself is retained with its hash and is not read when this page is built: the page and its evidence projection are a pure function of the admission artifact named above, so the same inputs always produce the same bytes.