Dell Enterprise Hub Update: On-Premises AI Models and Applications

Dell and Hugging Face have announced a new version of the Dell Enterprise Hub, providing a comprehensive toolkit for deploying generative AI models and applications on-premises. This update enables enterprises to roll out secure, in-network AI capabilities on Dell AI servers and AI PCs, reducing deployment timelines from weeks to under an hour.

Optimized Model Deployment for Dell AI Servers

The Dell Enterprise Hub provides fully tested containers optimized for specific Dell AI Server platforms, allowing for deployment and training via Docker and Kubernetes. This approach ensures that models are benchmarked and configured for maximum out-of-the-box performance.

Key available models include:

  • Meta Llama 4 Maverick (deployable on NVIDIA H200 or AMD MI300X Dell PowerEdge servers)
  • DeepSeek R1
  • Google Gemma 3

To ensure rapid availability, the hub works with Dell CTIO and Engineering teams to optimize new releases; for example, Llama 4 models were available on the hub within one hour of their public release by Meta.

Ready-to-Deploy AI Application Catalog

Beyond raw models, the Dell Enterprise Hub now includes an Application Catalog for deploying open-source AI applications within a private network. These applications utilize customizable helm charts to ensure Model Context Protocol (MCP) servers are registered during deployment.

Featured applications include:

  • OpenWebUI: Enables the deployment of on-premises chatbot assistants that utilize MCP to connect to internal data and services, supporting RAG (Retrieval-Augmented Generation) use cases via vector databases and web search.
  • AnythingLLM: A platform for building agentic assistants that connect to multiple MCP servers. It supports multiple models, image and document processing, and role-based access controls for internal users.

Multi-Hardware Accelerator Support

The Dell Enterprise Hub is the only platform providing ready-to-use model deployment solutions across the three major AI accelerator hardware providers:

  • NVIDIA: H100 and H200 GPU powered Dell platforms.
  • AMD: MI300X powered Dell platforms.
  • Intel: Gaudi 3 powered Dell platforms.

On-Device AI for Dell AI PCs

The hub has expanded to support on-device inference for Dell AI PCs powered by Intel or Qualcomm NPUs. This allows IT organizations to provide employees with private, on-device AI capabilities managed through Dell Pro AI Studio and fleet management systems like Microsoft Intune.

Supported on-device capabilities include:

  • Speech transcription: Powered by OpenAI Whisper.
  • Chat assistants: Powered by Microsoft Phi and Qwen 2.5.
  • Image upscaling and embedding generation.

Developer Tooling: CLI and Python SDK

To integrate the Dell Enterprise Hub into development workflows, Hugging Face has released the dell-ai open-source library. This library provides a Python SDK and a Command Line Interface (CLI), allowing developers to interact with the hub directly from their terminal or code via pip install dell-ai.

Sources