Praxis One Puts Robot AI and Motion Control on One Chip

Huaruizhipu unveiled Praxis One during China Robot Summit week in Hangzhou on September 17, 2026. Built around MediaTek’s Genio Pro 5100 system-on-chip, the on-robot computing platform is intended for heavy-duty humanoids and other robots that need substantial processing for perception and movement. 710

Why combine decision-making and control?

A robot may use one processor to interpret its surroundings and plan an action, then pass instructions to a separate motion controller. Huaruizhipu’s approach puts computing for both roles on one chip. The company says this could reduce cross-chip communication delays, data-transfer overhead and the effort required to make two hardware systems work together. Processing decisions on the robot can also avoid a cloud round trip for local tasks. 71017

That is the design case, not a measured result. The cited reports do not give a Praxis One control-loop time or establish that it beats a particular two-chip robot system in deployment. 717

What is it designed to run?

Praxis One is positioned for multimodal perception, vision-language-action (VLA) models—which connect what a robot sees and is told to the actions it takes—local language-model inference, lightweight world-model workloads and complex task planning. Reports also describe the Genio Pro 5100 CPU as suited to kinematics calculations and planning in changing environments. Together, these capabilities are meant to support a more direct path from sensing to movement decisions on high-degree-of-freedom robots. 7104

Reported Genio Pro 5100 specifications include a TSMC 3 nm process; eight Arm v9.2 CPU cores—one Cortex-X925, three Cortex-X4 and four Cortex-A720; 12 MB of L3 cache; 10 MB of system cache; and up to 260K DMIPS. Its eighth-generation NPU 890 is quoted at 106 TOPS and includes a Transformer-focused engine. Reports also quote local Qwen3-7B text generation at 23 tokens per second. 345

Those are chip specifications and a reported model speed, not evidence that all proposed workloads run at once or that motion control meets a particular latency target. Text-generation speed is not a motion-control benchmark. MediaTek separately cites over 50 TOPS of system-level generative-AI acceleration; that differently defined figure should not be treated as interchangeable with the reported 106 TOPS NPU rating. 1413

Praxis One, Praxis Nano and the wider stack

Praxis One is the higher-compute Genio Pro 5100 option for larger robots and heavier perception, inference and planning workloads. Praxis Nano uses the Genio 720 and focuses on real-time motion control, synchronized sensor collection and local feedback loops for smaller robots and lightweight data-collection devices. MediaTek rates the Genio 720 NPU at 9 TOPS, but chip ratings alone do not compare the finished platforms’ control performance. 71012

Huaruizhipu presented both platforms alongside its Ruimou Xingyuan data-collection ecosystem, a physical latent-space world model and robots for specific scenarios. Its proposed turnkey approach combines computing, software, data tools and robot applications; the cited coverage does not establish how much of that integrated offering has reached customers. 71031

Chinese coverage calls the overall platform “智能超脑”—literally “intelligent super-brain”—and names Praxis One and Praxis Nano as its two products. The available sources do not establish a definitive official English name for the overall platform. 710