Not All RTX Spark Laptops Are Created Equal

Not All RTX Spark Laptops Are Created Equal

More

Summary

NVIDIA’s RTX Spark laptops pair an ARM CPU with an NVIDIA GPU that share a single pool of memory, and they now ship from several vendors. Alex Ziskind visits Micro Center in Austin, Texas to compare four of them: the HP OmniBook X 14, the Microsoft Surface 15-inch, the MSI Prestige N16 Flip and the ASUS ProArt P16.

The video walks through screens, keyboards, trackpads, ports and weight, then narrows the field to two finalists to test back at the office. The key point for AI developers is memory. Conventional laptops with NVIDIA GPUs are limited to 8 GB on an RTX 5070 and 24 GB on an RTX 5090, while on 128 GB RTX Spark models a CUDA program can reach about 110 GB. That lets much larger local models fit, though the LPDDR5X memory is not as fast as dedicated GDDR7.

As a demonstration, gpt-oss 120B runs at about 65 tokens per second plugged in, and around 50 tokens per second with a 32,000-token context loaded. Ziskind also describes an automated battery test that repeats a simulated developer workday every half hour until the battery dies, and he includes the ROG Flow Z13 for comparison. The video closes with which laptop he picks as his daily driver and who should spend $3,000 to $4,000 or more on one.


📺 Source: Alex Ziskind · Published November 20, 2026
🏷️ Format: Comparison

1 Item

Channels