Flagship AI Mini PC
EVO-X2 AI Mini PC
AMD Ryzen™ AI Max+ 395
126 TOPS AI Compute Matrix
Built for local LLM deployment, AI development, image generation, machine learning workflows, and advanced AI computing. The EVO-X2 represents GMKtec's most powerful AI-focused Mini PC platform.
Local LLMs
AI Dev Platform
Image Gen
Neural Workflows
126 TOPS NPU
Processor (CPU)
- AMD Ryzen™ AI Max+ 395
- 16 cores / 32 threads
- TSMC 4nm FinFET process
- Up to 5.1 GHz (MAX)
- 16 MB L2 Cache / 64 MB L3 Cache
- TDP: Runs steadily at 120W (peak up to 140W)
Graphics & AI NPU
- GPU: Radeon 8060S Graphics (40-core RDNA 3.5 architecture)
- AI NPU: Based on XDNA 2 architecture
- 50 TOPS NPU Performance / Up to 126 TOPS overall
- LM Studio: AI performance 2.2× higher than NVIDIA RTX 4090 with lower power consumption
Memory & Storage
- RAM: Onboard LPDDR5X (non-upgradeable) | 64GB or 128GB options @ 8000MHz
- Primary ROM: PCIe 4.0 M.2 2280 SSD (2TB or 4TB capacities)
- Storage Expansion: Dual M.2 2280 slots (Max total 16TB, up to 8TB per slot via PCIe GEN4 x4)
- Note: SATA drives are not supported
Connectivity & I/O
- Video Output: HDMI 2.1, DisplayPort 1.4, USB4 (Max 7680×4320 @ 60Hz)
- Network: WiFi 7 (RZ717 MT7925) + 1 × RJ45 2.5G Gigabit LAN
- Bluetooth: Version 5.4
- Power: DC 19.5V, 11.8A (~230.1W) | AC 100–240V
Front & Rear Panels
- Front: 2x USB-A 3.2 Gen2, 1x Audio Combo, 1x USB Type-C (USB4.0), 1x SD Card Reader (SD4.0 SDXC)
- Front Buttons: Power, System Fan Lighting, Performance Mode Switch
- Rear: 1x DP 1.4, 1x HDMI 2.1, 1x USB-A 3.2 Gen2, 2x USB-A 2.0, 1x USB-C (USB4.0), 1x Audio, 1x DC IN, 1x RJ45 2.5G LAN
OS, Form Factor & LLM Support
- OS: Windows 11 Pro / Ubuntu
- Chassis Dimensions: 193 mm x 185.8 mm x 77 mm
- Finish: B/C-grade metal with CNC sandblasted oxidized finish; A-shell plastic spray coating
- 64GB Layout: DeepSeek-R1: 32B
- 96GB Layout: DeepSeek-R1: 32B, GPT-OSS: 120B
- 128GB Layout: DeepSeek-R1: 32B / 70B, Llama 4: 109B, Qwen3: 235B
128G Version - Local AI Performance Benchmark
| Model | Client Token Time | Avg. Token Speed |
|---|---|---|
| Deepseek-R1: 32B | 10.50s | 9.26 tokens/s |
| OpenAI GPT-OSS: 120B | 0.53s | 19.25 tokens/s |
| OpenAI GPT-OSS: 20B | 0.11s | 57.15 tokens/s |
| Qwen3: 30B | 0.18s | 55.02 tokens/s |
| Qwen3: 235B | 0.03s | 11 tokens/s |
| Llama4: 109B-Q4KM | 0.02s | 13.83 tokens/s |
| QWQ: 32B | 0.30s | 9.65 tokens/s |
EVO-X2 and T1 Model Compatibility
| Model Type | Model Name | EVO-X2 128G | EVO-X2 64G | EVO-X2 96G | EVO-T1 |
|---|---|---|---|---|---|
| Deepseek | Deepseek-R1: 14B | ||||
| Deepseek | Pre-installed Deepseek-R1: 32B | ||||
| Deepseek | Deepseek-R1: 70B | ||||
| Qwen | Qwen2.5: 7B | ||||
| Qwen | Qwen2.5: 14B | ||||
| Qwen | Qwen3: 30B | ||||
| Qwen | Qwen2.5: 72B | ||||
| Llama3 | Llama3: 8B | ||||
| MiniCP | MiniCP: PM-V | ||||
| QWQ | QWQ: 32B | ||||
| Llama4 | Pre-installed Llama4: 109B-Q4KM | ||||
| OpenAI | GPT-OSS: 20B | Not Supported | Not Supported | ||
| OpenAI | GPT-OSS: 120B | Not Supported | |||
| Intel Custom | Deepseek-R1: 32B Super |