Humanoid Robots Beat Usain Bolt’s 100m Record at Beijing Games—But the Real Test Is Work

The second World Humanoid Robot Games opened in Beijing with a headline built for viral clips: Tiangong Ultra ran 100 meters in 9.39 seconds, faster than Usain Bolt’s 9.58-second men’s world record. Honor’s Lightning also finished under that mark at 9.47 seconds. 13

Those results should not be treated as new human-athletics records. They were robot-event performances in a different competitive setting, and reports described unstable finishes, including robots losing balance or colliding after crossing the line. The more meaningful story is the combination of rapid technical progress, a much larger competition and the industry’s attempt to prove that humanoids can do more than perform for an audience.

Tiangong Ultra and Lightning’s sprint results

Tiangong Ultra, developed by the Beijing Humanoid Robot Innovation Center, won its preliminary heat after recovering from an early deficit and passing Honor’s Lightning near the finish. Its 9.39-second time was a dramatic improvement on the 21.50 seconds it recorded when it won the same event at the inaugural Games in 2025. 138

Lightning, developed by smartphone company Honor, placed second in the formal heat in 9.47 seconds. Before the Games, it had reportedly completed a preparatory run in 9.32 seconds and reached a peak speed of 14.5 meters per second. The 9.32-second run was a test performance, however, not the official opening-day heat result. 13

Performance	Time or mark	Context
Tiangong Ultra	9.39 seconds	Opening-day 100-meter preliminary heat 13
Honor Lightning	9.47 seconds	Same preliminary heat 38
Honor Lightning	9.32 seconds	Preparatory test; peak speed of 14.5 m/s 13
Usain Bolt	9.58 seconds	Human men’s 100-meter world record set in 2009 113
Tiangong Ultra	21.50 seconds	Winning time at the inaugural 2025 Games 38

The gap between the robots’ formal times and Bolt’s record is striking: Tiangong Ultra was 0.19 seconds faster, while Lightning was 0.11 seconds faster. But a robot crossing a marked distance faster than a human does not establish that it is a superior sprinter in the athletic sense. The machines have different bodies, control systems and operating constraints, and the reported balance problems underline how incomplete speed alone can be as a measure of performance.

A 2.88-meter jump—and the limits of robot “records”

The opening ceremony also featured a humanoid robot demonstration that reportedly reached 2.88 meters in the standing high jump. That is above the 2.45-meter human standing high-jump mark attributed to Javier Sotomayor in the event coverage. 5

The comparison is attention-grabbing but not equivalent to replacing a human athletics record. The available reports do not establish enough detail about the robot’s developer, measurement protocol or technical conditions to support a formal record comparison. It is best understood as a demonstration of robotic power, balance and control rather than a directly comparable sporting achievement.

The Games grew from a showcase into a broad test programme

The 2026 event runs for five days, from August 22 to 26, at Beijing’s National Speed Skating Oval. Organizers listed 666 teams and 2,056 robots from 16 countries across six continents. The inaugural 2025 Games had about 280 teams, more than 500 robots and 26 events.

This year’s programme was reported as 51 events and 1,301 matches, divided between conventional robot sports and scenario-based tasks. The sports programme includes disciplines such as sprinting, football, table tennis, martial arts, gymnastics, long jump, weightlifting and tug of war.

The scenario-based events are particularly important for judging commercial potential. They include tasks modeled on factory, retail, logistics and domestic work, including parcel sorting, mobile-phone packing, household assistance and related service activities. 17

That structure changes what the Games are trying to demonstrate. A sprint tests dynamic movement over a short distance. Sorting parcels or packing products tests perception, manipulation, endurance, repeatability and safe operation—capabilities that matter much more to a factory, warehouse or household.

Why the Beijing event matters beyond sport

The Games followed the World Robot Conference, which opened in Beijing on August 19. Together, the events created a concentrated showcase for Chinese robotics companies trying to move humanoids from impressive demonstrations toward commercial deployment. 717

The push is also part of China’s broader ambition to lead AI-powered humanoid robotics. Unitree CEO Wang Xingxing described the industry as approaching a possible “ChatGPT moment” for robot brains, reflecting the belief that improvements in AI could make machines more capable across many tasks.

At the same time, humanoid robots have become part of the broader U.S.–China technology rivalry. In July, the U.S. Federal Communications Commission added new foreign-made humanoid and quadruped robots to its Covered List, citing national-security concerns. The restriction targets new models that cannot obtain the required authorization, creating a potential obstacle for Chinese manufacturers’ future U.S. sales.

The result is a sector caught between spectacle and scrutiny. Fast runs and high jumps attract attention and signal progress in balance, actuation and control. They do not, by themselves, show that a robot can work an entire shift, recover from unexpected conditions, avoid damaging equipment or deliver value at a competitive cost.

The real benchmark is reliable useful work

The opening day showed that humanoid robots can produce remarkable headline numbers under controlled conditions. It also showed why those numbers need context: the race was not directly comparable with human athletics, and unstable finishes exposed the gap between maximum performance and dependable operation.

The next benchmark is therefore not another comparison with Bolt or Sotomayor. It is whether robots can safely and repeatedly sort, pack, carry, assist and respond in real-world environments. Industry observers and companies still face unresolved questions about reliability, cost and actual demand. 17

Beijing’s robot Games are valuable precisely because they make that contrast visible. The spectacle demonstrates what the machines can do once. The industrial and service events begin to test whether they can do it again—and keep doing it when the conditions are less forgiving.