Both models share the same AMD Ryzen 9 8945HX platform, 32 GB DDR5, 1 TB NVMe, and COOJ Sparrow MQ5 chassis. The AI-Accelerated model adds an NVIDIA A2000 GPU for faster inference and heavier workloads.
CPU-only configuration. Runs Gemma 4 E2B and E4B comfortably for everyday queries, system management, and lighter agentic tasks. Full Docker app marketplace, self-hosting, and XE network participation.
E2B (2.3B) and E4B (4.5B) run smoothly on CPU. 26B MoE is feasible at reduced speed. 31B Dense not recommended without GPU.
With NVIDIA A2000 GPU. Significantly faster inference for 26B MoE and 31B Dense models. GPU-accelerated media transcoding, enhanced multimodal processing, and future-proofing for heavier AI workloads.
All four models run with GPU acceleration. 26B MoE and 31B Dense deliver higher throughput and lower latency. Smooth 24/7 operation at full capability.
