XPeng's September: 41,256 Deliveries and the L03 Reaches Australia

XPeng has published its September delivery figures, and the headline number is the least interesting part of them. The group delivered 41,256 vehicles in September, up 5% month on month, taking the third quarter to 118,390, up 15% quarter on quarter — but down very slightly year on year against the 41,581 it delivered in September 2025, which was a record at the time.

What the model mix says

Three lines in the release matter more than the total.

The GX is holding above 7,000 a month. XPeng delivered 7,224 of its large SUV in September — the third consecutive month above 7,000, and the second consecutive month in the top three for large SUVs priced above ¥300,000. It also reports more than 3,000 orders in the Middle East. A ¥300,000-plus SUV selling at that rate is the part of the range that carries margin, and it is the answer to the "brand can't go upmarket" argument.

The MONA L03 passed 10,000 — and then 14,000. The group release gave a floor of over 10,000 for September with capacity still climbing; XPeng's model-specific post the following day put the actual figure at more than 14,000 (we covered the L03's September and its buyer profile here). The car is now on sale and delivering in Hong Kong, the Philippines, Australia and Indonesia. For readers here that is the significant line: the L03 is the model that precedes the G9L into this market, and Australian coverage of its rollout has already flagged software behaviour as the thing to watch.

The G9L launched at home. China-market launch on 17 September at a limited-time ¥231,800, with the car slated to reach 64 countries and regions — we covered the full pricing and specification stack here.

Software and robots, in the same release

September was also the month XPeng began pushing XOS 6.3.0 and the second-generation VLA Lite (our coverage of that rollout), and the month it activated an automated production line for high-order general-purpose humanoid robots, with IRON mass production targeted for the end of this year.

That last item is the one to treat with care. The "world's first" framing is XPeng's own, and the company's robot timeline — production by end-2026 with China deliveries from Q2 2027 — was set out at the G9L press conference (we covered those dates here). What the line does show is that the robot programme has moved from demonstrations to a manufacturing claim, which is a different kind of statement.

Why the +5% matters more than it looks

Sequential growth of 5% is modest. Set it against the market: Citi's read was that Chinese EV orders fell 22% week on week in September and that full-month retail was heading for a 7–12% year-on-year decline. In that context, a 5% monthly increase is not a company outrunning a boom — it is one holding volume while the market contracts, and doing it with an unusually large share coming from a ¥300,000-plus model and from export markets.

It also lands the day after a shareholder critique of XPeng's domestic weakness and investor communication was circulating on Weibo (our analysis of that thread is here). The release answers part of it — the domestic mix is improving at the top, and the L03 is now selling in four overseas markets — and leaves the rest open: the loss, the cost base, and whether the export growth translates into margin.

How the month compares

Set against the rest of the Chinese new-energy field, XPeng's September reads differently:

Brand	September 2026	Year on year
Leapmotor	105,656	+59%
XPeng	41,256	≈ −0.8%
NIO	37,408	+7.7%
Li Auto	31,817	−6.3%
Zeekr	37,216	+103.9%

The comparison comes from the Weibo post that prompted this update — and the full ten-brand league table for the month, with BYD's 463,600 at the top, is here. Its central observation is stark: NIO, XPeng and Li Auto together delivered 110,481 vehicles — barely more than Leapmotor's 105,656 on its own. A gap of 4,825 cars between three brands and one.

The mirror image is what each car earns. On the poster's estimate, derived from first-half financials, NIO averaged about ¥271,000 per vehicle, Li Auto ¥236,000 and XPeng ¥169,000, against roughly ¥100,000 for Leapmotor. His summary of the standoff is neat: the premium three want Leapmotor's volume, and Leapmotor wants their pricing.

Both halves of that are uncomfortable for XPeng, which sits in the middle. It grew 5% sequentially and slipped marginally year on year, in a month when Leapmotor grew 59% and Zeekr more than doubled. Volume leadership in this market is being decided below XPeng's average selling price, and premium pricing is being defended above it.

Caveats

• These are monthly wholesale delivery figures, the company's own and unaudited. They are not registrations, not retail sales, and a monthly number can include channel fill.
• The "top three above ¥300,000" ranking is XPeng's own characterisation and does not name the data source or the comparison set.
• "World's first" is a company claim about the robot line, as is the year-end mass-production target.
• The 64-country rollout is an intention, with no per-market timing in the release.
• Our figures come from XPeng's own post. An earlier version of our September coverage cited a third-party headline figure of 41,581 (+10.3%); that was XPeng's September 2025 number resurfacing in search results, not this year's. XPeng's September 2026 total is 41,256, and the quarter's three months reconcile with its stated Q3 figure of 118,390. The earlier article was corrected.
• The per-vehicle revenue figures are estimates, derived by the poster from first-half financials rather than disclosed by the companies. Mix differences — XPeng's export share, NIO's sub-brands, Li Auto's model changeover — make revenue-per-car comparisons rougher than they look.

The bottom line

A flat-ish month with a better mix, a global small car, and a robot line that is now a manufacturing claim rather than a video. XPeng's problem has never been whether it can build interesting things — it is whether the interesting things pay for themselves. The GX's run above 7,000 and the L03's arrival in four export markets are the first two lines in that argument; the loss in the next set of results is the counter-argument.

Source: 小鹏汽车 (Weibo), with market data from Citi via AAStocks — translated & adapted for XPENGTech.