• GMKTec EVO-X2 achieves faster generation speeds than DGX Spark across large models
• First-token latency improvements highlight the efficiency of AMD Ryzen AI Engine
• Mini PC form factor delivers workstation-level AI performance

GMKtec’s EVO-X2 has emerged as a strong alternative to the Nvidia DGX Spark, offering high-performance AI computing at roughly half the price.

The company says its compact mini PC delivered faster generation speeds for medium to large models such as GPT-OSS 20B and Llama 3.3 70B in recent benchmarks.

First-token latency was also lower across multiple tests, highlighting the efficiency of AMD’s Ryzen AI Engine and its optimized memory bandwidth.

GMKTec EVO-X2 excels for now

The EVO-X2 supports a wide range of open source models and frameworks, including Llama, Qwen, Mistral, Ollama, LM Studio, and vLLM, allowing developers to run AI workloads locally without relying on cloud resources, offering both flexibility and data privacy.

The device’s compact design also makes it suitable for high-performance mobile workstations where space and energy efficiency are important.

In terms of raw performance, the EVO-X2’s dedicated AI acceleration delivers multi-threaded, low-latency inference that competes directly with the DGX Spark.

The device handles real-world AI workloads effectively, enabling local deployment of large language models without the need for expensive enterprise infrastructure.

Its performance-to-power ratio is especially strong, providing workstation-level capability in a smaller, energy-efficient package.

GMKtec’s EVO-X2 is the world’s first mini PC to launch with AMD’s flagship Ryzen AI Max+ 395 APU.

The device is priced at €1,499 (approximately $1735), less than half of the Nvidia DGX Spark’s $3,999 cost.

This large price difference is hard to ignore, especially for organizations and individuals weighing budget against performance.

Despite this difference, the Nvidia DGX Spark remains a premium, enterprise-focused system, offering benefits in areas such as long-term reliability, multi-GPU scalability, and enterprise support.

Users with very large-scale or sustained workloads may still prefer Nvidia’s established system.

However, for developers, researchers, and professionals looking for a capable workstation or a flexible business laptop, the EVO-X2 presents a practical, lower-cost alternative.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.