Hangzhou, September 13–14, 2025 — GOSIM Hangzhou 2025, a premier conference spotlighting open-source and AI frontiers, officially kicked off at the Platinum Hanjue Hotel (Xihu District, Hangzhou). Organized by the Global Open-Source Innovation Meetup (GOSIM) and hosted by CSDN, the event positioned itself around “International Collaboration, Community Co-Creation, and Scenario-Based Interaction.” This year’s edition attracted more than 200 global industry leaders and over 1,500 developers.
As a Gold Sponsor, GMKtec, a pioneer in desktop AI supercomputing, brought a full lineup of flagship products to the stage, including the EVO-X2, EVO-T1, and EVO-X1, empowering developers and enterprises with powerful AI hardware.
About GOSIM (Global Open-Source Innovation Meetup)
GOSIM is a high-level international platform that gathers experts, developers, and enterprises from the open-source ecosystem. It is dedicated to fostering collaboration, innovation, and the practical application of open-source technologies. On this influential stage, GMKtec’s desktop AI supercomputers, known for their AI performance, portability, and adaptability, quickly became the focus of attention.
EVO-X2: Ushering in a New Era of Desktop AI Supercomputing
Powered by the AMD Ryzen AI Max+ 395 processor, built on the Zen 5 architecture with 4nm technology, the EVO-X2 boasts 16 cores and 32 threads, clock speeds up to 5.1GHz, and 80MB (L2+L3) cache.
-
Benchmark Performance: CINEBENCH R23 multi-core score of 36,516 pts and single-core score of 1,998 pts.
-
AI Power: Equipped with a 50 TOPS XDNA 2 NPU, the system achieves 126 TOPS total AI performance. In LM Studio, AI inference on a 70B dense model outperforms the NVIDIA GeForce RTX 4090 by 2.2×, while reducing power consumption by 87%.
-
Large Model Capability: Runs 200B+ parameter models locally, including Qwen-3 235B at 14 tokens/sec and OpenAI’s GPT-OSS-120B at 40 tokens/sec.
EVO-T1: A New Carrier for Open-Source AI Innovation
The EVO-T1 is powered by the Intel Core Ultra 9 285H processor, built on TSMC’s cutting-edge 3nm process. Featuring 16 cores and 24 threads (8 P-cores + 16 E-cores), with boost speeds up to 5.4GHz, it delivers strong hybrid performance.
-
AI Capability: Total peak performance of 99 TOPS (Int8), with 13 TOPS from the NPU.
-
Use Cases: Supports local deployment of models such as DeepSeek 32B, running at ~15 tokens/sec, enabling offline AI-powered applications such as creative design, voice transcription, and scientific data processing — secure, efficient, and independent from cloud reliance.
EVO-X1: The Compact Desktop AI Workstation
Equipped with the AMD Ryzen AI 9 HX370, built on Zen 5 architecture and 4nm process technology, the EVO-X1 features 12 cores, 24 threads, and 5.1GHz boost speeds.
-
Performance: Single-core score of 2,028 pts and multi-core score of 23,751 pts in CINEBENCH R23.
-
Graphics: Integrated Radeon 890M GPU (RDNA 3.5, 16 CUs @ 2.9GHz), outperforming many entry-level discrete GPUs. Oculink interface allows external GPU docking.
-
AI Power: Second-gen XDNA 2 NPU with 50 TOPS, delivering 80 TOPS combined system power (CPU+GPU+NPU) for AI workloads such as generative art and speech recognition.
GMKtec on Stage: Building the Foundation for Edge AI
At the event, Richard Han, GMKtec’s Channel Director, delivered a keynote speech highlighting how desktop AI supercomputers are the ideal carriers for edge AI deployment and a foundation for future infrastructure.
He emphasized that in the era of software transformation and AI compute decentralization, new hardware forms must meet requirements such as local deployment, data security, TCO optimization, and developer-friendly environments. GMKtec solutions address the needs of developers, research institutes, startups, SMEs, one-person companies, domain experts, and even home AI applications.
Live Demonstrations & Workshops
At the “Applications & Agents” forum and interactive demo zone, GMKtec showcased its desktop AI supercomputers through static displays and live demos:
-
Setting up open-source development environments on-site.
-
Real-time model deployment and optimization tutorials by GMKtec engineers.
-
Hands-on workshops where EVO devices served as lightweight yet powerful AI hardware.
-
Highlight demos included EVO-X2 paired with an RTX 4090 external GPU and developers coding on EVO-T1 in real projects.
The EVO-T1 was also selected as the official workstation for GOSIM 2025, supporting the entire event with its robust computing power.
This year’s agenda features 2 high-profile keynotes, 2 premium forums (the GOSIM AI Vision Forum and Open for SDG), 5 vertical-themed forums, 12 technical workshops, and 4 Hackathon innovation contests, along with a special 10th Anniversary celebration of Rust (RustGlobal + RustChinaConf). The participation of Intel and GMK not only brings top-tier hardware and technical support to the event but also leverages their brand influence to attract more ecosystem partners, fostering greater attention to open-source innovation and driving the collaborative development of global open-source and AI technologies.
Real-World EVO-X2 AI Applications
-
Enterprise AI — Local deployment for product R&D, sales insights, financial analysis, process optimization, legal consulting, smart customer service, and more.
-
Developers & Researchers — A reliable local AI platform for experimentation, model validation, and app development.
-
Content Creators — Performance suitable for video editing, 3D modeling, and creative workflows, while also serving as a home AI hub.
-
Gamers — Capable of running AAA games and AI-assisted gameplay for new immersive experiences.
Live Demo: EVO-X2 Connected to an External RTX 4090 GPU
Live Demo: Developers Using EVO-T1 for Project Development
Built-in Local Models: Ready Out of the Box
The EVO-X2 features pre-installed local AI models and knowledge bases, enabling users to enjoy instant, secure, and cloud-free AI experiences.
-
Comes with LLaMA 109B (128GB version) preloaded.
-
Supports Qwen3-235B and DeepSeek-70B locally with optimized scheduling.
-
Up to 128GB unified memory, with 96GB configurable as VRAM, ensuring smooth deployment of ultra-large models like Qwen3-235B and DeepSeek V3/R1 Distill-Llama-70B.
Recognition & Endorsements
-
Personally signed EVO-X2 by Dr. Lisa Su, AMD CEO.
-
Officially recommended by AMD.
-
Endorsed by Ji Zhaohui, VP of AMD Greater China Marketing.
About GMKtec
Shenzhen GMK Technology Co., Ltd. is a national high-tech enterprise specializing in desktop AI supercomputing centers and mini PCs. The brand GMKtec is a recognized “Shenzhen Famous Brand,” alongside Huawei and DJI.
-
Global leader in mini PCs and desktop AI supercomputing.
-
Strong R&D, production, and brand marketing capabilities.
-
Strategic partnerships with Intel, AMD, and Microsoft.
-
Products sold in 70+ countries with 35+ global distributors and service centers.
-
Established Longhua Industrial Base in 2019 for end-to-end development, testing, logistics, and global distribution.
GMKtec continues to drive innovation in edge AI computing, delivering powerful, compact, and versatile AI hardware that empowers enterprises, creators, and developers worldwide.
#AI #AIPC #AISupercomputing #MiniPC #AMD #Ryzen #AIMax+395 #RyzenMiniPC #innovation #AIhardware #EVO-X2 #EVO-T1 #EVO-X1 #GOSIM #GOSIMHangzhou2025 #desktopAI #Intel #Ultra9285H
