Tencent Robotics XChina2026

HY-Embodied-0.5

An AI model: a vision-language-action model (VLA) takes camera images and an instruction and turns them into movements for a robot. This page describes the software, not a robot.

Robotsnot stated

VersionsHY-Embodied-0.5 MoT-2BHY-Embodied-0.5-X

The developer has released this model in several versions. Each figure applies to the version its source names.

Key figures as stated by the developer

4 of 6 stated · 2 gaps

Model type ⁠A embodied reasoning
Controls not stated
Inputs ⁠A camera language
Access ⁠A open download
Parameters ⁠A 4,000,000,000
Base model not stated

How to read the figures

14kgA30.9lbA: The letter in the grey square after a figure points to the source the figure came from. Robots with several sources carry several letters.
74.5lb: Imperial figures are our own conversion unless the manufacturer states an imperial figure itself. The source letter always belongs to the metric figure — a conversion has no source of its own.

The four states

Versions and notes in the key figures (5)

Model type

Developer wording: "complex embodied reasoning (prediction, interaction, and planning)".

Controls

Note The model answers in text and with image coordinates, not with robot motion: "Coordinates are normalized to the integer range" 0 to 1000.

Inputs

The model takes images and text; the Hugging Face model card gives its task as "image-text-to-text".

Access

Tencent released HY-Embodied-0.5 "featuring the open-sourced" MoT-2B weights on Hugging Face; the extended variant has its own weights there. The licence does not cover the European Union, the United Kingdom or South Korea.

Parameters

Total for the open model: "The MoT-2B variant contains 4B total parameters", with "only 2.2B activated parameters" during inference. For the extended variant the model card says "architecture (4B total parameters with only 2B activated)". The 32B model of the family has no open weights on the model card.

Versions

HY-Embodied-0.5 MoT-2B: 4,000,000,000
HY-Embodied-0.5-X: 4,000,000,000

Full specifications

5 of 12 fields stated

The letter points to the source at the foot of the page. Fields without a figure are dashed: the developer does not state them — that is not the same as zero, and not the same as no.
Field	Value	Source and caveats
Input and output
Model type	`embodied reasoning`	A Developer wording: "complex embodied reasoning (prediction, interaction, and planning)".
Inputs	`camera` `language`	A The model takes images and text; the Hugging Face model card gives its task as "image-text-to-text".
Controls	not stated	Note The model answers in text and with image coordinates, not with robot motion: "Coordinates are normalized to the integer range" 0 to 1000. ⁠B
Degrees of freedom controlled	not stated
Size and speed
Parameters	4,000,000,000	A Total for the open model: "The MoT-2B variant contains 4B total parameters", with "only 2.2B activated parameters" during inference. For the extended variant the model card says "architecture (4B total parameters with only 2B activated)". The 32B model of the family has no open weights on the model card. Versions HY-Embodied-0.5 MoT-2B 4,000,000,000 HY-Embodied-0.5-X 4,000,000,000
Control rate	not stated
Where it runs and what it is built on
Runs where	not stated	Note Tencent describes the 2B model as made "for edge deployment" and lists "NVIDIA GPU with ≥ 16 GB VRAM" for the extended variant, but does not say that it runs on the robot. ⁠B
Base model	not stated
Openness
Access	open download	A Tencent released HY-Embodied-0.5 "featuring the open-sourced" MoT-2B weights on Hugging Face; the extended variant has its own weights there. The licence does not cover the European Union, the United Kingdom or South Korea.
License	Tencent HY Community License Agreement	C The licence file is the "TENCENT HY COMMUNITY LICENSE AGREEMENT", which says: "THIS LICENSE AGREEMENT DOES NOT APPLY IN THE EUROPEAN UNION, UNITED KINGDOM AND SOUTH KOREA". The weights of the extended variant carry the same Tencent licence; its code on GitHub is under Apache-2.0.
The developer’s own claims
Learns new tasks	not stated
Works across robots	not stated	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.

Notes

Tencent Robotics X and the HY Vision Team describe HY-Embodied-0.5 as a suite of foundation models for embodied intelligence, built for spatial-temporal perception and embodied reasoning.
The family has a 2B edge model and a 32B model; the open weights are the MoT-2B model.
HY-Embodied-0.5-X is an extended variant of HY-Embodied-0.5 with additional post-training, released on 24 April 2026 with its own weights and licence file.
Tencent says HY-Embodied-0.5 can act as the language-and-vision part of a vision-language-action model.

Sources

A Manufacturer product page. huggingface.co/tencent/HY-Embodied-0.5 · Retrieved 2026-10-13
B huggingface.co/tencent/HY-Embodied-0.5-X · Retrieved 2026-10-13
C huggingface.co/tencent/HY-Embodied-0.5/blob/main/LICENSE · Retrieved 2026-10-13

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.