RTX Spark mini PC promises to overcome memory limitations using Mechrevo’s unusual AI SSD

Notebookcheck reports that Chinese manufacturer Mechrevo is preparing the Master Road X10, a mini PC built around NVIDIA’s RTX Spark platform for local AI workloads. The platform can support up to 128GB of unified memory in its maximum configuration, but Mechrevo has not confirmed the exact memory and hardware configuration for this model.

The notable feature is Mechrevo’s proprietary AI SSD, which is intended to move portions of large models from DRAM to NAND flash and retain previously processed context between sessions. Mechrevo claims the design can support models with 80 billion, 120 billion, and 122 billion parameters, and cites earlier claims of up to 5.7 times more supported workload and a 95% faster prefill stage. Notebookcheck notes that detailed benchmarks, quantisation settings, and test conditions have not been published, while NAND is substantially slower than DRAM; these are manufacturer claims, not independent performance results.

For Canadian buyers, the report does not establish pricing, availability, warranty coverage, shipping, or a release date. The AI SSD capacity, connectivity, final memory configuration, and real-world inference performance also remain unknown. Canadian manufacturer listings and independent testing should be checked before treating the Master Road X10 or its storage-assisted design as proof of a particular model size or response speed.