57
Technology4h 2m ago

Local AI inference mini PCs, powered by AMD's Ryzen AI Max+ 395 chip, can now run 235B-parameter models on x86 architecture, potentially allowing developers to reduce or eliminate $440-per-month cloud inference subscriptions.

Archive Window: 7 Days Left

Shenzhen, China (GMKtec headquarters)

Who
AMD, GMKtec, Nvidia, Apple, HP, Lenovo
What
Local AI inference mini PCs, powered by AMD's Ryzen AI Max+ 395 chip, can now run 235B-parameter models on x86 architecture, potentially allowing developers to reduce or eliminate $440-per-month cloud inference subscriptions.
When
Sun, 14 Jun 2026 09:54:52 GMT · 4h 2m ago
Where
Shenzhen, China (GMKtec headquarters) ·
Why
The AMD Ryzen AI Max+ 395 chip integrates CPU, GPU, and NPU and shares 128 gigabytes of unified LPDDR5x-8000 memory, addressing the VRAM limitations of discrete graphics cards for large AI models.
The Frontline Impact

How this affects you

This development enables developers to locally run large AI models that previously required expensive cloud subscriptions, potentially saving substantial costs and enhancing data privacy. While suitable for many development and single-user tasks, these systems may not match the throughput of discrete GPUs for high-concurrency workloads.

Story chain

3 events in this thread

Verified Sources & Citations