Galbot 银河通用

Embodied-AI robot maker behind Galbot G1 and open GraspVLA model

Other Enterprise Open Source
Researched · Published · Reviewed
RECATOOLS Score
6.7 / 10
Capability
7
Value for money
6
Ease of use
5
ASEAN readiness
4
API quality
5
Founded
HQ
Users
Launched
Developer

Overview

Galbot (银河通用) is a Beijing embodied-AI startup founded in 2023 by Peking University's He Wang, building the dual-arm G1 mobile-manipulation robot and GraspVLA, an open-sourced grasping model trained on billions of simulated actions.

Advertisement

What you can produce with Galbot 银河通用

  • Galbot G1 dual-arm wheeled mobile-manipulation robot
  • GraspVLA open-source zero-shot grasping model (code + checkpoints on GitHub/Hugging Face)
  • Autonomous retail deployment (unmanned convenience stores)
  • OpenWBT open-source whole-body tracking codebase
  • Simulation playground and real-world robot control interface
  • 100+ published research papers per company materials
Advertisement

ASEAN Perspective

Galbot 银河通用 in Southeast Asia

ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).

RECATOOLS Verdict

Galbot's strongest signal isn't the G1 robot itself — it's that the company open-sourced GraspVLA, the grasping model behind it, complete with code, checkpoints and a simulation playground on GitHub. That's unusual restraint for a Chinese robotics startup burning through funding rounds, and it's earned genuine traction: the GraspVLA repo and sibling projects have picked up real stars and forks from the robotics research community, not just press coverage.

The G1 itself is a wheeled dual-arm robot doing autonomous retail work (unmanned convenience stores), which is a more grounded deployment than most humanoid demos. But funding velocity here is dizzying — multiple rounds inside three months in early 2026 pushing toward 5 billion yuan raised, mostly from state-linked funds. That's a bet on a two-year-old company's research pipeline, not proven commercial scale, and buyers should read it that way.

Independent AI-assisted assessment by RECATOOLS.

What people say

Galbot (银河通用, 'Galaxy General') was founded in May 2023 by Peking University professor He Wang, and it has raised capital at a pace few embodied-AI startups match: The Robot Report tracked a $300 million round to scale mobile-manipulator deployments, and by early 2026 the company had closed roughly 2.5 billion yuan in a single new round — pushing total funding across just three months to nearly 5 billion yuan, with backers including the National Artificial Intelligence Industrial Investment Fund, Sinopec (China Petrochemical), China Investment Corporation and SAIC Finance. One tracker puts total funding since founding at around $800 million as of January 2026.

The clearest evidence of technical substance, rather than hype, is GraspVLA: an end-to-end grasping model pre-trained on SynGrasp-1B, a billion-scale synthetic action dataset, published at CoRL 2025 and open-sourced on GitHub in July 2025 with model checkpoints, a simulation playground and a real-world control interface. It's a genuine zero-shot result — the model grasps object categories it never saw training demonstrations for — and shipping working code, rather than a paper-only claim, is what separates Galbot from robotics startups that only release demo videos.

Galbot's GitHub organization backs that up: OpenWBT (whole-body tracking) and OpenTrack have each pulled 700+ stars, and a follow-on project, Humanoid-GPT, was accepted at CVPR 2026. Numbers like that come from other labs actually using the code, not from marketing.

On the product side, the G1 — a wheeled, dual-arm mobile manipulator first shown publicly in March 2025 — is running in autonomous convenience-store deployments in China, a more concrete commercial application than most humanoid-robot demos manage. There's no public pricing, no self-serve access and no independent commercial review trail (G2/Capterra don't cover industrial robot suppliers); what public sentiment exists is research-community reception of the open-sourced models, which has been genuinely positive by GitHub star counts and pickup, rather than customer testimonials in the traditional sense.

Summary of public user & expert reviews, compiled by RECATOOLS.

About this listing

Researched on
Published on
Last reviewed

This entry was compiled from publicly available data including Galbot 银河通用's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Galbot 银河通用 unless explicitly stated.

Data accuracy

Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.

For the latest details, please refer to Galbot 银河通用 directly →

Spotted something out of date? Suggest an update →

Advertisement