Owners of Intel Mac Pro machines with AMD video cards can still run local AI projects, according to AppleInsider. The ToshLLM project says it offers native Metal acceleration, zero cloud dependence, and no per-token billing.
The project's own claim is that it runs without sending data to remote servers and without the usage fees that come with many cloud AI services. Users keep their workloads on the hardware they already own.
Performance will not match newer systems. The article notes that the setup will not be as fast as the M5 Ultra Mac Studio, but it lets older hardware take on local AI work instead of sitting idle.
The ToshLLM project is available at toshllm.com. The original AppleInsider report covers the setup in more detail.
Source: AppleInsider · Summarized by HeadlinesBriefing