Framework has officially unveiled its first desktop PC, a machine engineered specifically to handle high-end gaming and local AI inference tasks, with preorders now open and shipping scheduled for early Q3 2025.
The Controversial Memory Trade-off
While modularity is the company’s hallmark, this new desktop introduces a divisive design choice: the RAM is soldered directly onto the board. Users can choose between 32GB and 128GB configurations at the time of purchase, but they will not be able to upgrade or replace the memory later.
“There is one place we did have to step away from PC norms, though, which is on memory. To enable the massive 256GB/s memory bandwidth that Ryzen AI Max delivers, the LPDDR5x is soldered,” Framework CEO Nirav Patel wrote on the company’s blog.
“We spent months working with AMD to explore ways around this, but ultimately determined that it wasn’t technically feasible to land modular memory at high throughput with the 256-bit memory bus,” he added.
Powering Local AI Workloads
Despite the lack of RAM modularity, the inclusion of up to 128GB of unified memory creates significant opportunities for users interested in large language models. The system is capable of running Llama 3.3 70B seamlessly using Ollama, llama.cpp, and various open-source tools designed for local AI processing.
Beyond Llama, other open-weight models from providers like Mistral, Nous, Hermes, or DeepSeek are expected to run efficiently. For power users, Framework is also offering the mainboard as a standalone component without a case, enabling setups like a custom mini-rack with four mainboards running in parallel for intensive AI testing.
Pricing and Software Compatibility
The base model of the Framework Desktop is priced at $1,099, with the premium configuration reaching $1,999. In line with the company’s commitment to software freedom, the system supports Windows as well as major Linux distributions, including Ubuntu, Fedora, and the gaming-optimized Bazzite.
Leave a Reply