MagicLabChina2026
Magic-VLA K02
Key figures as stated by the developer
1 of 7 key figures stated-
Model type A
vision-language-action
Not stated: Controls · Inputs · Access · Parameters · Base model · Licence
Full specifications
2 of 14 fields stated| Field | Value | Source and caveats |
|---|---|---|
| Release | ||
| Released | 18 July 2026 | [A] "2026-07-18 21:01" The Paper reports that MagicLab showed the model at WAIC 2026 on 18 July; MagicLab's own article calls its IFA 2026 showing the model's overseas debut. B |
| Input and output | ||
| Model type | vision-language-action | Developer wording: "Magic-VLA K02, MagicLab's proprietary vision-language-action model". A |
| Inputs | not stated | |
| Controls | not stated | |
| Degrees of freedom controlled | not stated | |
| Size and speed | ||
| Parameters | not stated | |
| Control rate | not stated | |
| Where it runs and what it is built on | ||
| On robot or cloud | not stated | |
| Base model | not stated | Note MagicLab describes the architecture: "Magic-VLA K02 uses a dual-system hierarchical architecture: a high-level planner that breaks tasks into sub-goals and generates target reference images, paired with a low-level vision-language-model backbone and dynamic expert modules for smooth, continuous motion." A |
| Openness | ||
| Access | not stated | |
| Licence terms | not stated | |
| Licence | not stated | |
| The developer’s own claims | ||
| Learns new tasks | not stated | |
| Works across robots | not stated | Note The article names no robot models that run Magic-VLA K02. A Yes only if the developer itself says the model runs on different robot models, for example robots from two manufacturers or two of one manufacturer’s models. Several units of the same robot do not count. |
How to read the figures
- 50HzA
- The letter in the grey square after a figure points to the source the figure came from. Pages with several sources carry several letters.
- 50HzB
- A dashed letter marks a secondary source – a datasheet, manual or developer documentation from the manufacturer, or a press report used only where the manufacturer gives no figure.
Notes
- MagicLab describes Magic-VLA K02 as its proprietary vision-language-action model.
- MagicLab lists Magiclab Robotics Technology (Wuxi) Co., Ltd. as the company behind its site.
Versions and notes in the key figures (2)
Model type
Developer wording: "Magic-VLA K02, MagicLab's proprietary vision-language-action model".
Base model
Note MagicLab describes the architecture: "Magic-VLA K02 uses a dual-system hierarchical architecture: a high-level planner that breaks tasks into sub-goals and generates target reference images, paired with a low-level vision-language-model backbone and dynamic expert modules for smooth, continuous motion."
Sources
-
A
Primary source. www.
magiclabglobal. · Retrievedcom/ magiclab-brings-physical-ai-to-europes-factory-floors-at-ifa-2026-debuting-full-stack-embodied-intelligence-stack. html -
B
Secondary: www.
thepaper. · Retrievedcn/ newsDetail_ forward_ 33613775
One line per source, not per figure: a source has one retrieval date. The letter beside a figure points to the line it came from.