Apollo Go (Apollo RT6)

The world’s second-largest driverless ride service — and the only one that has drawn a written protest from the drivers

By paid fully driverless volume, Apollo Go sits behind only Waymo: 3.2 million rides in Q1 2026, weekly peaks above 350,000 in March, more than 22 million cumulative public rides, 27 cities. This catalog already carried Waymo One, WeRide’s GXR and Zoox, and had a hole where the number two should be. This entry fills it.

What earns it its own page, though, is not the ride count. It is that Apollo Go is the only autonomous service anywhere with a documented labor dispute attached to it.

The vehicle

The Apollo RT6 is Baidu’s sixth generation, contract-built by Jiangling Motors. Baidu’s 2022 specification lists 38 sensors including 8 LiDAR and 12 cameras, dual automotive-grade compute units delivering 1,200 TOPS with full hardware and software redundancy, and a detachable steering wheel. The production car measures 4,765 mm long on a 2,830 mm wheelbase, uses a 110 kW BYD FinDreams motor and a swappable LFP battery with a roughly three-minute swap, tops out at 135 km/h, and seats four.

Two things need saying plainly. On LiDAR count, Baidu’s own spec says eight while on-site reporting from the May 2024 production launch said five, with 40 sensors total; the discrepancy has never been reconciled. On the steering wheel, the design is detachable but delivered vehicles still have one, pending Chinese regulation. Battery capacity and range have never been disclosed.

As for a seventh generation: there is no Baidu announcement and no deployment, only one unattributed rumor. The RT6 is the car carrying the rides.

Cost, fare, and why they are not the same number

The widely-cited RMB 204,600 is the per-vehicle build cost Baidu showed on an invoice at its May 2024 production event. It is not a price. Baidu does not retail the RT6; it operates the vehicles or supplies them to partners — Lyft owns the operational value chain in Europe while Baidu provides vehicles, and Uber’s agreement covers markets outside the US and mainland China. No updated cost figure exists for 2026.

The fare is the more useful number. In Wuhan in 2024, a 5 km Apollo Go ride ran about 7.5 yuan against 11-15 yuan on Didi and roughly 15 yuan for a metered taxi. But a reporter’s own metered 6 km comparison came out at 9.38 yuan for Apollo Go against 7.54 yuan for ordinary ride-hail and about 15 yuan for a taxi. The honest conclusion: it consistently undercuts the metered taxi, and does not consistently undercut a discounted ride-hail trip. There is no published 2025-2026 fare data, so anyone telling you what it costs today is guessing.

What happened on the drivers’ side

In late June 2024, Wuhan Jianshe Automotive Passenger Transportation filed a petition with the Wuhan municipal transport authority asking it to limit Apollo Go’s operations, on the grounds that robotaxis were “taking jobs from the grass roots.” Four of the company’s 159 drivers had quit since April that year citing lost income. Over the same period the city’s transport website logged more than 300 traffic complaints about Apollo Go — lane positioning, slow reaction at lights, sudden braking.

The petition matters not because it worked — there is no evidence Wuhan capped permits in response — but because it is the only documented written protest filed by the displaced side of any robotaxi rollout anywhere.

It also shows where the ceiling on substitution sits. The Ministry of Transport’s November 2023 interim guideline requires a remote safety operator ratio no lower than 1:3, and Apollo Go’s Wuhan operation runs at exactly 1:3 — one cloud-cabin operator watching three vehicles. Reporting from a Wuhan operating site in August 2024 described a team of 40-plus people running 20-plus vehicles: cloud safety operators, in-car safety officers, dispatchers, ground crew, engineers and admin. Many of the cloud operators were redeployed in-car safety officers and experienced former ride-hail drivers.

One ratio worth getting right

Chinese ride-hail driver earnings really are collapsing, and Apollo Go is not why. Ministry of Transport data shows 979 million ride-hail orders nationwide in June 2026 across 401 licensed platforms. Caixin reports average driver income of 27 yuan per hour in 2024, down 12.9%, against roughly 18% growth in driver and vehicle supply. Licensed driver certificates reached about 7.48 million by 2024, and only 10.5% of those drivers contribute to basic pension and medical insurance.

Set that against Apollo Go’s roughly 3.3 million rides per quarter versus China’s roughly 979 million ride-hail orders per month — about one tenth of one percent. Anyone attributing the income collapse of Chinese ride-hail drivers to robotaxis has the causation backwards. What crushed those earnings was the explosive growth in the number of licensed human drivers.

Why we care for LostJobs

The value of this entry is that it breaks “robots are taking the jobs” into three checkable facts instead of one feeling.

First, the displacement is real but legally capped. A 1:3 operator ratio means that under current Chinese rules, one driverless vehicle eliminates at most two-thirds of a driver’s job. Any model that counts one vehicle as one unemployed driver overstates the effect threefold.

Second, there is a genuine path from the displaced job to the created one, and it is narrow. Cloud-cabin operators were drawn from in-car safety officers and veteran ride-hail drivers — a rare case of the displaced actually walking into the replacement role. But a 40-person team running 20 vehicles is an early-stage staffing ratio that must compress as the fleet grows. Do not read it as a steady state.

Third, and most easily missed: administrative risk moves your job security more than technical progress does. When roughly 100-200 vehicles stopped at once in Wuhan on 31 March 2026 and stranded passengers for up to two hours, three ministries summoned eight AV companies within weeks and China froze new robotaxi permits nationwide until a late-June review. For that stretch, Wuhan’s taxi drivers were protected not by the limits of the technology but by a software failure. If you drive a taxi or a ride-hail car in a tier-one Chinese city, the thing to watch over the next two to three years is not Baidu’s product launches. It is the operator-ratio rule and the pace of permit issuance.