Hugging Face has launched an inference-as-a-service product for AI deployment on NVIDIA's DGX Cloud. This service leverages NVIDIA NIM microservices to enhance token efficiency and provide access to popular AI models for developers.

The new service will deliver up to five times better token efficiency, enable immediate access to NVIDIA NIM microservices, and support leading AI models like Llama 3 and Mistral AI. Developers can prototype and deploy open-source AI models from the Hugging Face Hub, benefiting from serverless inference, increased flexibility, minimal infrastructure overhead, and optimized performance with NVIDIA NIM on the NVIDIA DGX Cloud.

Analyst QuickTake: Hugging Face offers integrated MLOps solutions, with a platform akin to GitHub for AI code repositories, models, and datasets. Launching its inference-as-a-service capabilities for foundation models such as LLama 3 and Mistral AI expands its capabilities into LLMOps solutions.