A machine shaped like a human, moving faster than any human can.
In Shanghai, a machine shaped like a human has crossed a threshold that once belonged only to living bodies: MirrorMe Technology's Bolt humanoid robot reached 10 metres per second in real-world conditions, becoming the fastest bipedal robot ever tested outside a laboratory. The achievement is less about a number than about a direction — China has positioned itself at the frontier of robotic athletics, and the ambition behind Bolt reaches beyond trophies toward the training floors of elite sport. In a decade defined by questions about what machines can do alongside humans, Bolt offers one answer: they can outrun us, and then help us run faster.
- A red humanoid robot in Shanghai just broke the speed barrier for bipedal machines, hitting 10 m/s in real-world testing — a feat no full-size humanoid had achieved outside a controlled lab.
- The company behind Bolt, MirrorMe Technology, has been quietly building toward this moment since 2016, and their quadruped Black Panther II already reportedly outpaced Boston Dynamics' WildCat on a live national broadcast.
- The race is not just mechanical — Beijing hosted the first World Humanoid Robot Games last August, signaling that robotic athletics has become a domain of national competition and strategic investment.
- MirrorMe is now positioning Bolt not as a record-holder but as a 'steel sparring partner' for elite athletes, a training tool that moves faster than any human competitor ever could.
- The trajectory is clear: speed was the proof of concept, but the destination is a robot precise and powerful enough to serve as a performance multiplier in professional sport.
A humanoid robot named Bolt, finished in red metal and standing 175 centimetres tall, recently ran at 10 metres per second on a treadmill in Shanghai — making it the fastest bipedal robot ever tested outside a laboratory. The machine belongs to MirrorMe Technology, a Shanghai-based firm whose researchers have been working on high-speed humanoid locomotion since 2016, though the company itself was only formally founded in May 2024.
The engineering behind Bolt is invisible to the eye but decisive in effect: redesigned joints and a power system built to replicate human movement while sustaining speeds no human can match. In a demonstration, founder Wang Hongtao ran alongside Bolt on parallel treadmills. The robot's strides were shorter, but its cadence was so rapid it left him behind without apparent effort.
Before Bolt, MirrorMe built Black Panther II, a quadruped that sprinted 100 metres in 13.17 seconds on a Wuhan athletics track during a live CCTV broadcast — a performance that reportedly surpassed Boston Dynamics' WildCat. The speed record itself is secondary to what it represents: China competing at, and winning, the frontier of robotic athletics.
MirrorMe frames Bolt not as a trophy machine but as a 'super-species' robot — one capable of approaching or exceeding human athletic ability. The practical vision is a 'steel sparring partner' for Chinese athletes, a training tool that pushes competitors beyond what any human rival could demand of them. This ambition sits within a broader national context: last August, Beijing hosted the first World Humanoid Robot Games, where a humanoid ran 100 metres in 21.5 seconds. Speed was always the measure for MirrorMe. It was never the only goal.
A humanoid robot in red metallic finish stood on a treadmill in Shanghai, moving its legs in a blur. Its name is Bolt, and it just became the fastest bipedal robot ever tested outside a laboratory, hitting 10 metres per second in real-world conditions. The achievement belongs to MirrorMe Technology, a Chinese robotics firm that has spent the last decade chasing a single obsession: how fast can a machine shaped like a human actually run.
Bolt stands 175 centimetres tall and weighs 75 kilograms—proportions the company describes as the humanoid robot's "ideal form." But the real engineering lives in what you cannot see: newly designed joints and a power system built to move like a human body while sustaining speeds no human can match. In a promotional demonstration, the company's founder Wang Hongtao raced alongside Bolt on treadmills. The robot took shorter strides than the man but compensated with a cadence so rapid it left him behind. This is not a laboratory trick. This is a full-size machine running at genuine speed in the real world.
MirrorMe was only formally established in May 2024, but the team behind it has been working on this problem since 2016. The company is based in Shanghai with its core researchers drawn from Zhejiang University. Before Bolt, they built Black Panther II, a quadruped that in 2025 exceeded 10 metres per second and then, in a live broadcast on China Central Television, sprinted 100 metres in 13.17 seconds on a Wuhan athletics track. That performance reportedly surpassed Boston Dynamics' WildCat, which research papers from 2020 clocked at approximately 8.8 metres per second. The speed record itself matters less than what it signals: China is now competing at the frontier of robotic athletics, and it is winning.
The company's ambition extends beyond the trophy case. MirrorMe frames Bolt not as a speed-record machine but as what it calls a "super-species" robot—a machine capable of approaching or even exceeding human athletic abilities through advanced motion perception and physical performance. The practical application is striking: the company envisions Bolt as a "steel sparring partner" for Chinese athletes, a training tool that can move at speeds beyond what human competitors can achieve, pushing athletes to new limits. This is not science fiction. It is a deliberate strategy to use robotics as a performance multiplier in sport.
The context matters. Last August, Beijing hosted the first World Humanoid Robot Games, where the Tien Kung humanoid won the 100-metre race in 21.5 seconds. These are not niche events. They signal a national commitment to robotics as both a technological frontier and a domain of athletic competition. MirrorMe's product lineup already includes Baobao and Apollo alongside Black Panther II, each designed for different applications. Bolt represents the next step: a humanoid robot fast enough to train elite athletes, precise enough to replicate human movement, and capable of sustaining extreme speeds in real conditions. The company has made clear that speed was never the only goal. But it was always the measure.
Citações Notáveis
The company aims to develop 'super-species' robots capable of approaching or surpassing human athletic abilities through advanced motion perception and physical performance.— MirrorMe Technology
Bolt is envisioned as a 'steel sparring partner' for Chinese athletes, using human-level perception and athletic skill to help them push beyond real-world speed limits.— MirrorMe Technology (via Auto News)
A Conversa do Hearth Outra perspectiva sobre a história
Why does a Chinese robotics company care so much about making robots run fast? What's the actual use case?
Speed is the proof. If you can build a bipedal machine that moves at 10 metres per second in the real world, you've solved problems in balance, power delivery, and motion control that apply to everything else—walking, climbing, working in unpredictable environments. The speed record is the headline, but the engineering underneath is what matters.
But you mentioned this "steel sparring partner" idea. Is that real, or marketing?
It's real enough to be worth taking seriously. An athlete training for the 100-metre dash needs to run against something faster than themselves to improve. A human coach can't do that. A robot that can sustain 10 metres per second consistently, without fatigue, becomes a tool. Whether it works as well as the company claims is an open question, but the logic is sound.
How does this compare to what Boston Dynamics is doing?
Boston Dynamics has been focused on mobility and dexterity—robots that can navigate rough terrain, climb, manipulate objects. MirrorMe is laser-focused on one thing: bipedal speed. It's a different philosophy. Boston Dynamics' WildCat maxed out around 8.8 metres per second. Bolt hit 10. That's not a huge gap, but in a competitive space, it's enough to matter.
The company was only founded in 2024. How did they move this fast?
The team has been working on this since 2016, before the company even existed. They came from Zhejiang University. So you have a decade of research suddenly crystallized into a startup with a clear mission and the resources to execute. That's why the speed of iteration looks so fast—it's not actually fast, it's just compressed.
What happens next? Is this the ceiling, or can they go faster?
Almost certainly faster. Every speed record in robotics has been broken within a few years. The real question is whether the practical applications—training, research, industrial use—catch up to the speed benchmarks. Right now, Bolt is a demonstration of capability. Making it useful is the next phase.