Framework deployed the Framework Desktop in February 2025, a compact mini PC built and explicitly marketed for running large AI models — like those that power chatbots — directly on a user's desk, without sending data to a cloud service.

The Desktop uses AMD's Ryzen AI Max processors with a wide memory bus that makes large amounts of RAM accessible to its graphics chip, supporting up to 96 GB of graphics-addressable memory. This makes it well-suited for local AI inference — running large language models (LLMs) on personal hardware rather than through an online service. Framework's product page specifically highlights running models like Llama 70B as a use case. The company also promotes connecting multiple Desktops together over fast networking to run even larger models such as DeepSeek R1 671B. AMD gave away 100 Framework Desktop units to AI developers at launch to accelerate ecosystem development.