AMD released AIMs, standardized AI inference microservices, on September 30, 2026, saying they deliver a unified user experience across Radeon, Instinct, and EPYC hardware. It is the company's first software update for AI deployment since the Multi-Accelerator Support blog in 2025.
AMD reported AIMs support deployment on Radeon PRO R9700, EPYC 9965, and Instinct MI300X hardware, measured through containerized workflows. That compares with earlier AI deployment methods requiring manual configuration.
AIMs are built on Docker and ROCm, targeting large-scale model serving with reduced deployment complexity. Availability begins with public Docker images, initially for developers and enterprise users.
"AIMs abstract away the complexities involved in configuring and serving AI models by providing a mechanism that automatically selects runtime parameters based on the user’s input, hardware, and model specifications," said Rasmus Larsson, AMD’s AI lead. The system dynamically selects profiles based on model, precision, and hardware.
The announcement follows the Multi-Accelerator Support blog, which introduced AIMs as part of AMD’s AI solution blueprints. AMD did not say how AIMs will scale with future hardware, and the company raised the question of how users will customize deployment beyond built-in profiles.
Source: amd
