NVIDIA brings agents to life with DGX Spark and Reachy Mini
NVIDIA is integrating its DGX Spark computing platform with the Reachy Mini robot to enable embodied AI agents. The collaboration, highlighted on the Hugging Face blog, demonstrates running agent workloads on edge hardware for robotics applications. This represents a convergence of NVIDIA's inference infrastructure with open robotics platforms.
Related guides (4)
Related events (8)
Reachy Mini - The Open-Source Robot for Today's and Tomorrow's AI Builders
Hugging Face has published a blog post introducing Reachy Mini, an open-source desktop robot designed for AI developers and researchers. The post positions the robot as a platform for building and testing embodied AI applications. As an open-source hardware/software project, it targets the growing intersection of robotics and AI model deployment.
Reachy Mini goes fully local
A Hugging Face blog post describes running the Reachy Mini robot's conversational AI stack entirely on local hardware, eliminating cloud dependencies. The post likely covers the models, tooling, and inference setup required to achieve on-device operation for a small consumer robot. This represents a deployment case study at the intersection of edge inference and robotics.
Serverless Inference with Hugging Face and NVIDIA NIM
Hugging Face and NVIDIA have partnered to offer serverless inference via NVIDIA NIM microservices on DGX Cloud infrastructure. The integration allows developers to run optimized model inference without managing GPU infrastructure, combining Hugging Face's model hub with NVIDIA's inference optimization stack. This represents an expansion of the existing Hugging Face–NVIDIA partnership into managed inference services.
NVIDIA's GTC 2025 Announcement for Physical AI Developers: New Open Models and Datasets
NVIDIA announced new open models and datasets for physical AI development at GTC 2025, covered via the Hugging Face blog. The release targets robotics and embodied AI developers with open-weights resources. This represents NVIDIA's continued push into the physical AI ecosystem alongside its hardware dominance.
Easily Train Models with H100 GPUs on NVIDIA DGX Cloud
Hugging Face announced integration with NVIDIA DGX Cloud, enabling users to train models on H100 GPU clusters directly through the Hugging Face platform. The partnership simplifies access to high-end training infrastructure without requiring users to manage cloud provisioning themselves. This represents a continued push to lower the barrier to large-scale model training for the broader ML community.
Hugging Face integrates MCP tools with Reachy Mini robot
Hugging Face published a blog post describing how to add Model Context Protocol (MCP) tools to the Reachy Mini robot platform. The integration connects MCP-based tool-calling infrastructure to physical robotics hardware. This is a concrete deployment example of MCP expanding beyond software agents into embodied AI systems.
DeepLearning.AI launches Context Hub for coding agents; Google releases Nano Banana 2 image generator
Andrew Ng and collaborators released Context Hub (chub), an open CLI tool that provides coding agents with up-to-date API documentation to reduce hallucinated or outdated API calls. Google separately launched Nano Banana 2 (Gemini 3.1 Flash Image), a faster and cheaper image-generation system built on Gemini 3 Flash's mixture-of-experts architecture, priced at roughly half its predecessor and claiming the top spot on Arena.ai's text-to-image leaderboard. The newsletter also references Claude Opus 4.6 as a leading coding model and notes the growth of agent-to-agent social infrastructure (OpenClaw, Moltbook) as context for the tooling need.
Strands Agents and LeRobot enable direct deployment from Hugging Face Hub to robot hardware
A Hugging Face blog post describes an integration between Amazon's Strands Agents framework and the LeRobot robotics library, enabling models from the Hugging Face Hub to be deployed directly onto physical robot hardware. The post demonstrates a pipeline connecting cloud-hosted model weights to real-world robotic control. This is relevant to the growing agent-tool ecosystem and the practical deployment of embodied AI.