- GMKtec EVO-T2 mini PC reaches 180 TOPS using combined CPU, GPU and NPU acceleration
- Its PCIe 5.0 storage introduces data speeds greater than 10 GB per second
- On-premises AI models run without relying on external cloud infrastructure
At a recent launch event, GMKtec introduced GMKtec EVO-T2, a compact desktop system built for local AI computing.
According to the company, the device integrates third-generation Intel Core Ultra processors and claims to have up to 180 TOPS of computing capacity.
It combines CPU, GPU and NPU resources and enables local execution of large language models with up to 70 billion parameters without relying on external cloud infrastructure.
Article continues below.
Computer architecture and AI workloads
The EVO-T2 is based on Intel’s Panther Lake architecture and is manufactured using the 18A process node, incorporating RibbonFET transistors and rear power delivery.
These design elements are associated with higher efficiency and transistor density, although most of the performance data referenced remains tied to internal benchmarks.
The company claims that complex workloads such as code generation and document processing can be quickly executed with this device.
For some tasks, GMKtec says the EVO-T2 completes them in seconds under controlled conditions.
Graphics capabilities are handled by the integrated Intel Arc B390 GPU, which includes twelve Xe cores and support for DirectX 12 Ultimate, real-time ray tracing, and AI-assisted scaling.
This setup allows the system to extend beyond AI inference to areas such as rendering and visual content workflows.
Despite its small size, the device includes two M.2 storage slots that support PCIe 5.0 and PCIe 4.0, with a total capacity of up to 16TB.
In theory, PCIe 5.0 SSDs are capable of sequential speeds exceeding 10 GB/s, with some exceeding 15 GB/s, while PCIe 4.0 drives typically reach around 7 GB/s under optimal conditions.
For connectivity, it includes USB4 with 40Gbps bandwidth and OCuLink support for external GPUs.
Additionally, the system supports dual Ethernet configurations, offering 10GbE and 2.5GbE networking.
To address memory limitations, Phison collaborated with GMKtec to integrate aiDAPTIV+ AI SSD technology.
This system dynamically expands available memory by distributing workloads between DRAM and storage, allowing large models to be segmented at runtime.
The active parts are processed on the GPU, while the less active data remains stored in the memory and SSD layers.
This “pseudo-memory” mechanism is described as reducing bottlenecks when processing large models.
However, its implications for long-term performance under sustained workloads have not been independently verified.
GMKtec claims that it “effectively overcomes traditional DRAM limitations,” a claim that may require independent validation.
The system ships with a pre-configured AI environment, allowing immediate access to AI tools and models without manual configuration.
This is another OpenClaw-compatible Chinese device, similar to the Minisforum N5 Max, despite warnings from the Chinese government about its security vulnerabilities.
OpenClaw allows EVO-T2 to run autonomous AI agents locally, performing tasks from data processing to content generation without relying on cloud services.
However, the software has known weaknesses that can be exploited to steal sensitive data and has previously been used to distribute malware through GitHub repositories.
Microsoft has even advised against running OpenClaw on standard personal or enterprise devices, leaving the wisdom of GMKtec’s decision in question.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds. Be sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp also.
