Red Hat said that it will acquire Neural Magic, which specializes in generative AI inference workloads.
Terms of the deal weren't disclosed.
For Red Hat, a unit of IBM, Neural Magic will enable the company to align its open platform to AI workloads. Red Hat said it is planning to make generative AI more accessible to enterprises via vLLM, an open source project for serving multiple models. Neural Magic has been a big backer of the vLLM project.
In a statement, Red Hat said it will combine Neural Magic's focus on vLLM with its hybrid cloud AI technologies to run open source models, fine tune LLMs and improve inference performance. Neural Magic has a vLLM-based stack with infrastructure choice, security policies and model lifecycle management and offers a unified library for optimizing LLMs.
Red Hat AI includes Red Hat Enterprise Linux AI, a foundation model platform, Red Hat OpenShift AI, a platform to train models across Kubernetes environments, an InstructLab, which fine tunes IBM's Granite LLMs.
Neural Magic's two primary products are Nm-vllm, an enterprise inference server for LLMs on GPUs, and DeepSparse, an inference server for models on CPUs.
- Snowflake launches Arctic LLM, to release details under Apache 2.0 license
- IBM open sources Granite models, integrates watsonx.governance with AWS SageMaker
- IBM rolls out Granite 3 models, makes default for Consulting Advantage
- Meta's Zuckerberg lays out AI vision with heavy dose of open source
- Nvidia drops NVLM 1.0 LLM family, revving open-source AI
