LingBot-VLA
$ 0
An open robot brain from Ant Group’s Robbyant that turns images and plain-language instructions into arm actions, trained on about 20,000 hours of real robot data.
Available on backorder
Brain Score
2
Specifications and details:
| Nationality | China |
|---|---|
| Website | https://www.robbyant.com |
| Model type | Foundation Model |
| Manufacturer | Robbyant (Ant Lingbo Technology), part of Ant Group (China) |
| Release date | January 27, 2026 |
Description
LingBot-VLA is an open vision-language-action model from Robbyant, the embodied AI company of Ant Group. The team calls it a pragmatic foundation model, so the focus sits on real robots rather than lab demos. It learned from about 20,000 hours of real-world manipulation data. That data came from nine popular dual-arm robot setups. Robbyant also found that performance kept climbing as the data grew, with no sign of a plateau. As a result, the release adds strong evidence that robot learning benefits from scale.
The model turns camera views and plain-language instructions into arm and gripper actions. For example, it can clear and place tableware after a short round of fine-tuning. An optional depth version also helps robots handle transparent objects such as glass vases. Developers can download two 4B checkpoints from Hugging Face and ModelScope under the Apache-2.0 license. In addition, Robbyant ships a fast post-training toolkit, so teams can adapt the model to their own robots. LingBot-VLA anchors the company’s LingBot family and its goal of one brain for all robots.
New! 2026 Humanoid
Robot Market Report
198 pages of exclusive insight from global robotics experts — uncover funding trends, technology challenges, leading manufacturers, supply chain shifts, and surveys and forecasts on future humanoid applications.
now Google DeepMind
Contact Humanoid.guide
Website: https://technology.robbyant.com/lingbot-vla
