# Ant Group's LingBot-VA 2.0: a robot brain that runs on one GPU

> Ant Group released LingBot-VA 2.0, a robot control model it says hits 93.6% in simulation on one GPU.

*Ant's robotics unit unveiled what it calls the first embodied-native world action model, claiming 93.6% success in simulation and single-GPU deployment.*

By The FeaturedDaily Desk · FeaturedDaily
Canonical: https://featureddaily.com/news/ant-lingbot-va-2-embodied-world-action-model-brief

> **Key:** **The news:** Ant Group's robotics unit Ant Lingbo released **LingBot-VA 2.0**, which it bills as the world's first "embodied-native world action model" for robots, claiming a **93.6% success rate in simulation** and operation on a **single GPU**.

## What it is

Most frontier models are trained on text and images and reason *about* the world. A world action model is trained to *output actions* -- motor commands and manipulation plans -- based on what a robot sees, so it can directly control hardware. "Embodied-native" means built for that from the ground up, not a language model bolted onto a robot arm. Ant says LingBot-VA 2.0 is the first of its kind, a framing that is itself part of the pitch rather than an independently settled fact.

## The demo and the numbers

In demos the model handles delicate objects -- holding potato chips without crushing them -- and performs everyday manipulation such as tidying a desk, the kind of soft, force-sensitive tasks that are notoriously hard for robots. The two figures Ant leads with: a **93.6% success rate in simulation tests**, and the claim that it **runs on a single GPU**, positioning it for low-cost, wide deployment. That efficiency angle is the real story, because it targets the cost that usually keeps capable robot control out of everyday machines. The launch lands in a busy week for embodied AI, alongside Rhoda AI's FutureVision and Mecka AI's robot-action-data business, continuing China's efficiency-and-deployment focus.

> Ant Lingbo, Ant Group's robotics unit, released LingBot-VA 2.0, described as the first "embodied-native world action model," reporting a 93.6% success rate in simulation, delicate-object manipulation, and operation on a single GPU.
> — [AITNT](https://m.aitntnews.com/ainews/m/en/date/2026-07-10), 2026-07-10

## The catch

> **Note:** The "world's first," 93.6% and single-GPU claims are **Ant's own, largely from simulation** -- no independent third-party benchmark is cited. Sim-to-real transfer is the hard part, and a chip-holding clip is not a deployed robot workforce. Treat the numbers as vendor-reported until verified.

## What's next

Watch for independent, real-hardware benchmarks -- not simulation -- and whether the single-GPU claim holds under real robot workloads with unpredictable objects and lighting. If it does, cheap on-device robot control is the genuinely disruptive part, because it lowers the cost of putting capable manipulation on ordinary machines. If it doesn't, this stays a promising demo among many that landed this same week.

## Key takeaways

- Ant Group's robotics unit Ant Lingbo released LingBot-VA 2.0, pitched as the first embodied-native world action model, one built to output physical actions rather than only reason about text.
- Ant reports a 93.6% success rate in simulation and demos of delicate manipulation such as holding potato chips without crushing them and organizing a desk.
- The headline hook is efficiency: Ant says the model runs on a single GPU, aimed at making capable robot control cheap enough to deploy at scale.
- The world's-first, 93.6% and single-GPU claims are Ant's own, largely from simulation, with no independent benchmark cited, so treat them as vendor figures until verified.

## FAQ

### What is a world action model?
A model trained to output physical actions -- motor commands and manipulation plans -- based on what a robot sees, so it can control hardware directly, rather than only reasoning about text or images.

### Are the 93.6% and single-GPU numbers verified?
No. They are Ant's own figures, largely from simulation, with no independent third-party benchmark cited. Sim-to-real transfer is unproven, so treat them as vendor claims.

### Does this mean robots are ready for homes and factories?
Not yet. The demos are controlled and the metrics are simulated. A chip-holding clip is a long way from a reliable, deployed robot workforce.

## Sources

- [Global AI News Daily — 2026.07.10](https://m.aitntnews.com/ainews/m/en/date/2026-07-10) — AITNT, 2026-07-10
- [The Latest AI News and Breakthroughs That Matter Most](https://www.crescendo.ai/news/latest-ai-news-and-updates) — Crescendo AI, 2026-07-10
- [AI News for the Week of July 10](https://solutionsreview.com/ai-news-for-the-week-of-july-10-updates-from-accenture-google-cloud-supermicro-more/) — Solutions Review, 2026-07-10