Itstone Zhihang debuts AWE3.0, A1 robot sets record

Chinese embodied intelligence company Itstone Zhihang unveiled its AWE3.0 general‑purpose embodied model at the AWE 2026 show in Shanghai on March 14, and used the event to spotlight a Guinness World Records achievement by its A1 robot. The robot completed 105 sub‑millimeter wiring‑harness assemblies in one hour during an industrial demonstration at the expo. The company positioned the model release and the record as evidence that embodied AI in China is moving from simulation and demos toward real, high‑precision factory work.

The Guinness‑certified test focused on a concrete manufacturing task: wiring‑harness assembly, a process that requires deformable cables to be inserted and aligned with sub‑millimeter accuracy. Itstone Zhihang said the A1 robot hit 105 assemblies in 60 minutes, a throughput figure the company used to show both precision and endurance. A Guinness record adds a third‑party benchmark to a field where many announcements rely on lab demonstrations or controlled videos.

AWE3.0 is described by the company as a model built for real‑world manipulation rather than simulation‑only tasks. In its launch materials, Itstone Zhihang highlighted sub‑millimeter precision, flexible object perception and control, long‑horizon task stability, and cross‑scene generalization as core capabilities. Those claims align with the use of a factory‑like assembly test at the expo, emphasizing a focus on tasks that require repeated accuracy rather than one‑off stunts.

The company also introduced SenseHub, a data acquisition kit designed to capture the multimodal signals needed to train embodied models. SenseHub was shown at AWE 2026 and won an AWE Innovation Award, according to the company’s event materials. The pairing of a new data tool and a model release signals an emphasis on data pipelines, which have become central to embodied AI progress in China’s robotics ecosystem.

Data collection is increasingly viewed as a bottleneck for embodied AI, since models need large volumes of real‑world interaction traces rather than only synthetic simulations. By packaging SenseHub as a product alongside AWE3.0, Itstone Zhihang is signaling that it wants to standardize how robots capture and reuse multimodal data on production floors. That framing matters for manufacturers that need a clear data workflow before they scale pilots beyond a single line or workcell.

The choice of wiring‑harness assembly as the record task is telling because it is a common and difficult industrial process. Harnesses are flexible, change shape during handling, and require consistent positioning under tight tolerance, which makes them a challenging target for automation. A claim of sub‑millimeter accuracy backed by a Guinness record is therefore aimed at the kind of operations managers who care about repeatable precision and throughput in real production lines.

The March 12–14 AWE 2026 expo provided timing and visibility for the launch. AWE is one of China’s largest consumer electronics and appliance shows, and this year’s event created a high‑traffic showcase for industrial AI as well. Itstone Zhihang’s March 14 launch event for AWE3.0 and SenseHub leveraged the show’s audience to frame the company as a bridge between cutting‑edge AI models and practical manufacturing adoption.

Market data suggests why companies are racing to prove real‑world capabilities. IDC forecasts that China’s embodied intelligent robot user spending will exceed $1.4 billion in 2025 and reach $77 billion by 2030, implying a 94% compound annual growth rate. Those projections point to a short window in which vendors can establish credibility and win pilot deployments before the market scales.

The combination of a model launch, a Guinness record and a data‑collection product is a strategic signal to the market. Itstone Zhihang is telling prospective partners that it can provide a full stack: the model (AWE3.0), the robot platform (A1) and the data pipeline (SenseHub). If the company can translate the record into stable factory trials, it could strengthen the case for embodied AI systems to move from pilots into repeatable, high‑volume tasks.

What changed with the March 14 announcement is that Itstone Zhihang now has a public, third‑party‑verified milestone tied to a concrete industrial process, rather than only model claims. What happens next will depend on whether the A1 and AWE3.0 system can sustain the same precision and throughput in longer‑term factory deployments, and whether manufacturers are willing to integrate the system into production lines at scale. If those pilots succeed, the company’s approach could become a template for how China’s embodied AI vendors move from demos to commercial rollout.