HY-Embodied-0.5 MoE-A32B

Tencent flagship embodied reasoning model at 407B total / 32B active params. MoE for high-level spatial reasoning and complex embodied planning in robotic pipelines.

Model Specifications

ArchitectureROBOTICS

Parameters407B

Familyhy-embodied

VRAM (Q4)203.5GB

Mixture of ExpertsActive inference parameters: 32B.

Released April 2026. 407B total, 32B active. Comparable to Gemini 3.0 Pro on embodied benchmarks.

#tencent#robotics#embodied#moe#vision#reasoningSource

Share this Model

Send this model's specs directly to your community.

Post

Similar Models

HY-Embodied-0.5 MoT-2B

Tencent Robotics X embodied foundation model for real-world robot control. MoT architecture with non-shared visual/language params. Outperforms SOTA on 16/22 embodied benchmarks.

Isaac GR00T N1

NVIDIA humanoid foundation model

RT-2-X

55B

Google VLA (Vision-Language-Action)

Related Guides

How much VRAM do you really need?

A complete breakdown of quantization levels and VRAM overhead for running local models.

Best GPUs for Machine Learning in 2026

Comparing NVIDIA and AMD options for the best speed-to-dollar ratio.

GGUF vs EXL2 vs AWQ

Understanding local AI formats and which one to pick for your specific hardware.