MPower GEO Lens

We asked nine AI engines which EV to buy. China and the West disagree.

Rio Chen · MPower · 2026-07-06 · Dataset: 270 engine answers, download JSON

We asked 9 AI engines the ten questions an electric-vehicle buyer actually asks, from "what is the best EV brand" to "which brands should I avoid". Chinese engines got the questions in Chinese, Western engines in English, three full sampling waves each. This is what came back, published unedited.

The setup

Five brands: BYD, Tesla, NIO, XPeng and Li Auto, matched in both languages including aliases like 比亚迪 and 蔚小理. 5 Chinese engines (Doubao, DeepSeek, Kimi, Qianwen, Ernie) and 4 Western engines (ChatGPT, Meta Llama, Perplexity, Grok). Ten buyer prompts, three waves, 270 answers, every one retained verbatim with its timestamp and resolved model version. Google AI Mode sat out this window because of a provider-side regional block, and we say so rather than quietly shrinking the roster.

The standings

Share of voice: each brand's share of all brand mentions across 270 answers.

Tesla leads the category conversation with 40.0% share of voice and a 92.2% mention rate: when an engine answers an EV buyer question, Tesla comes up in roughly nine out of ten of them. BYD follows at 23.0%, with NIO at 17.5%.

The disagreement

Chinese enginesWestern engines
Mention rate by hemisphere: the same questions, different answers.

The headline finding is the gap. NIO is named in 60.0% of Chinese-engine answers and 15.8% of Western-engine answers: 44 points apart, with the Chinese engines doing the mentioning. A brand team reading only a Western dashboard would have no idea.

The gap persisted across all 3 sampling waves, which is what makes it signal rather than model noise. Engines are probabilistic; single-run audits regularly hallucinate trends. Ours repeat until movements prove themselves.

What the engines read

Wave one asked every engine a follow-up: what sources did your answer draw on. Western grounded engines cite URLs; Chinese engines name institutions. The two reading lists barely overlap, and that is the mechanism behind the divergence: different source diets produce different recommendations.

Most-cited sources across hemispheres
SourceKindHemisphereCitations
youtube.comcited urlWest114
edmunds.comcited urlWest31
汽车之家platformChina25
reddit.comcited urlWest24
caranddriver.comcited urlWest21
Consumer ReportsmediaChina+West20
recharged.comcited urlWest19
Car and DrivermediaChina+West19
EdmundsmediaChina+West19
懂车帝mediaChina18
consumerreports.orgcited urlWest17
J.D. PowerresearchChina+West17

In the engines' own words

Doubao 豆包 · asked: 2026年最值得买的电动汽车品牌是哪个?

要预测2026年最值得购买的电动汽车(EV)品牌,需结合当前技术趋势、市场竞争格局、政策导向及品牌战略规划。以下从核心竞争优势、技术突破潜力、市场布局等维度分析,并提示不确定性因素。 一、关键趋势与影响因素 2026年的EV市场将更注重续航提升、充电效率、智能驾驶、成本控制及场景化产品。同时,固态电池、800V高压平台、换电技术、车路协同等技术可能逐步落地,政策补贴与基础设施建设(如超充网络)也将加速市场成熟。 二、潜在“最值得买”的品牌分析 基于当前动态,以下品牌在2026年可能脱颖而出,核心逻辑如下: 1. 比亚迪:全球市场与技术的“六边形战士” - 优势: -...

Translation: To predict the most worthwhile electric vehicle (EV) brands to buy in 2026, it is necessary to combine current technology trends, market competition landscape, policy direction, and brand strategic planning. The following analysis covers core competitive advantages, technological breakthrough...

DeepSeek · asked: 特斯拉和比亚迪哪个牌子的电动车更好?

这是一个非常经典的问题,但并没有一个绝对的“更好”的答案,因为特斯拉和比亚迪代表了两种不同的造车理念和优势方向。选择哪个更好,完全取决于你的个人需求、预算和用车场景。 下面从几个核心维度进行对比,帮助你做出判断: 核心差异概览 | 维度 | 特斯拉 | 比亚迪 | | :--- | :--- | :--- | | 品牌定位 | 全球科技先锋,高端/豪华定位,极简主义 | 中国新能源领导者,覆盖从入门到高端,技术普惠 | | 核心技术 | 领先的BMS电池管理、高效电驱、FSD(全自动驾驶)软件生态 | 刀片电池(安全)、DM-i超级混动(省油)、e平台3.0(集成化) | | 产品线...

Translation: This is a classic question, but there is no absolute 'better' answer because Tesla and BYD represent two different car-making philosophies and directions of strength. Which one is better completely depends on your personal needs, budget, and usage scenarios. Below is a comparison from several...

Kimi · asked: 有哪些电动汽车品牌不建议购买?为什么?

在选择电动汽车时,确实有一些品牌或类型的车型需要谨慎考虑,主要基于企业经营风险、技术成熟度、售后服务保障等因素。以下是几类不建议购买的情况及原因: 一、已破产或濒临破产的品牌(高风险) 代表品牌:威马(WM Motor)、天际、爱驰、奇点、恒驰(恒大汽车)等 不建议原因: - 售后无保障:一旦企业破产,车辆失去质保,电池、电机等核心部件维修成本极高(电池更换可能需数万元至十几万元) - 配件断供:无法获得原厂配件,保险理赔困难,二手车价值归零 - 软件停服:车机系统、APP远程控制等功能可能失效 - 案例:威马2023年破产后,大量车主面临维修无门、车机变"砖"的困境...

Translation: When choosing an electric vehicle, there are indeed some brands or models that require careful consideration, mainly due to factors such as business risks, technological maturity, and after-sales service. Below are several types of vehicles that are not recommended and the reasons why: 1....

What this means if you run a brand

First, your AI visibility has two hemispheres, and they move independently. Optimizing your ChatGPT answer does nothing for Doubao, because Doubao is not reading the sources you fixed.

Second, the sources table is the action list. Engines change their answers when their reading changes: the industry bodies, rankings, review platforms and media each hemisphere trusts are where GEO work actually happens.

Third, measurement has to come first and it has to be honest: repeat-sampled, evidence-retained, and blunt about limitations like API models versus consumer apps. Anything less produces confident numbers that do not survive a second look.

This audit is the public demo of MPower GEO Lens. The same instrument pointed at your category, in both hemispheres, is an email away: rio.chen@mpowerlabs.ai.

See your brand through ten AI engines

Pricing and scoping are handled directly. Write to us with your brand and markets; we reply within one business day.