GMKtec Featured Products
GMKtec is dedicated to redefining AI PCs, gaming PCs, office mini PCs, and creative computing experiences.
Featured in Media
GMKtec mini PCs have been featured in hands-on reviews, product roundups, and editorial coverage from respected tech publications worldwide.
GMKtec Is Recommended By
GMKtec has been widely reviewed and recommended by leading global tech media and creators, including Lon TV, ETA PRIME, Digital Foundry, ServeTheHome, Gizmochina, NASCompares, TechTablets, Iceberg Tech, and more. These trusted voices highlight GMKtec’s performance, reliability, and innovation in mini PCs and compact computing solutions.
Featured Blog
Meet GMKtec at GITEX GLOBAL 2026: Bringing 320B AI to Every Desk in the Gulf
Discover desktop AI supercomputing, next-generation Agentic PCs, and high-performance mini PCs at Booth H37-C44 in Dubai, December 8–11, 2026. As the Middle East accelerates its investment in artificial intelligence, an important question is emerging: How can businesses harness powerful AI capabilities while keeping sensitive data local, secure, and under their control? At GITEX GLOBAL 2026, GMKtec will showcase how desktop AI supercomputing can help bring advanced AI capabilities closer to where data is created and used. We invite technology enthusiasts, business users, distributors, channel partners, developers, and media representatives to visit our booth, explore our latest products, and exchange ideas about the future of AI computing. Bringing 320B-Parameter AI to the Desktop At the heart of our showcase is the GMKtec EVO-X5 Pro, our flagship desktop AI supercomputer powered by the AMD Ryzen™ AI Max+ PRO 495 processor and supporting up to 192GB of unified memory. First unveiled at IFA 2026 in a joint launch with Michael Nordquist, CVP, Client BU at AMD, the EVO-X5 Pro represents our vision of bringing advanced AI computing from the cloud to the desktop. With support for fully offline inference of large language models with up to 320 billion parameters under appropriate model, software, and system configurations, the EVO-X5 Pro opens new possibilities for developers, professionals, and businesses exploring local AI applications. As the UAE and Saudi Arabia advance their national AI ambitions, desktop AI computing can complement cloud platforms and data centers, offering another approach to deploying AI while maintaining greater control over data and workloads. Steven, CEO of GMKtec, said: “We believe the future of AI should not be limited by access to the cloud. With the EVO-X5 Pro, we are bringing powerful AI computing directly to the desktop, helping businesses and developers explore advanced AI workflows while maintaining greater control over their data. We look forward to connecting with partners and users across the Middle East to explore the opportunities ahead.” Explore Our Latest Computing Portfolio Beyond the EVO-X5 Pro, visitors can get an up-close look at a broader range of GMKtec products, including: ● NEO-X1 Pro: Combining an AMD Ryzen™ 9 9955HX3D processor with NVIDIA GeForce RTX 5070 graphics for high-performance gaming and graphics-intensive applications. ● EVO-X3: A powerful compact AI computing system designed for demanding workloads. ● EVO-T2S: Another addition to our evolving portfolio of intelligent computing solutions. From local AI computing and professional productivity to gaming and everyday applications, our G, M, K, EVO, and NEO series reflect our commitment to making high-performance computing more compact, accessible, and practical. From AI PCs to Agentic PCs AI is evolving beyond answering prompts and running individual applications. The next generation of computing will increasingly focus on understanding context, reasoning through tasks, and taking action on behalf of users. This is the vision behind the Agentic PC. GMKtec is working to bring this vision closer to everyday users through compact systems that support increasingly capable local AI workloads. At GITEX, we look forward to discussing how these technologies may support new workflows, business applications, and opportunities for collaboration across the Middle East. Let's Connect at GITEX GLOBAL 2026 Whether you are an existing GMKtec customer, a potential distributor, a channel partner, a developer, or simply interested in the future of AI computing, we warmly invite you to visit us. Come meet our team, explore our products up close, discuss your requirements, and exchange ideas about potential partnerships and opportunities in the region. We look forward to meeting familiar faces and making new connections across the Middle East and beyond. Event Details ● Event: GITEX GLOBAL 2026 ● Dates: December 8–11, 2026 ● Booth: H37-C44 ● Location: Expo City Dubai, UAE Mark your calendar and join us in Dubai as we explore the next generation of desktop AI supercomputing and Agentic PC innovation. See you at GITEX GLOBAL 2026!
Read moreSame Chip, Different AI Speed: How GMKtec EVO-X5 Pro Reaches 75 Tokens per Second
If you’ve been following the world of local LLM inference, you may have heard a common criticism of AMD’s Strix Halo platform: with 256 GB/s of memory bandwidth, it can run large language models, but don’t expect spectacular speeds There’s some truth to that—but only half the story. Hardware sets the limits. Software determines how much of that potential you can actually unlock. On the GMKtec EVO-X5 Pro, a specialized inference engine called Halogen demonstrates just how much difference software optimization can make. Running the same model on the same underlying hardware, Halogen delivers a substantial performance boost over a general-purpose inference engine. Here’s what makes the difference. 1. The Numbers: 23 Tokens/s vs. 75 Tokens/s Let’s start with the results. Running Qwen3.8-Flash-Next on the AMD Strix Halo platform, we compared the performance of a general-purpose inference engine with Halogen on the GMKtec EVO-X5 Pro. Inference Engine Sustained Generation Speed llama.cpp, general-purpose engine Around 23 tokens/s Halogen on GMKtec EVO-X5 Pro Around 75 tokens/s That’s more than 3× the sustained token generation speed under the tested conditions. And the gap can become even more pronounced when working with long contexts. In practical terms, tokens per second (tok/s) measures how quickly an AI model generates its response. At 75 tok/s, the model can produce text at a pace that feels much closer to a real-time conversation, making it easier to work through coding tasks, analyze documents, and build local AI workflows without constantly waiting for the next sentence. The key point? Both configurations use the same underlying chip. The difference comes from how the software puts that hardware to work. Note: These figures reflect GMKtec's internal testing under the specified configuration. Actual performance varies with model settings, context length, quantization, and workload. 2. The Secret Isn’t New Hardware. It’s a Different Engine. First, let’s clear something up: AMD provides the hardware. GMKtec hasn’t modified the silicon or changed the physical memory bandwidth. The real variable is Halogen, a specialized inference engine designed to optimize AI workloads on AMD’s Strix Halo platform. Unlike general-purpose engines built to support a wide range of hardware and model architectures, Halogen takes a much narrower approach. It focuses on a specific GPU architecture and a specific model family to extract more performance from the hardware. That specialization is the foundation of its advantage. 3. Built for One Platform, Not Every Platform General-purpose inference engines are designed to work across a broad ecosystem. They need to support different GPU vendors, hardware generations, model architectures, and execution paths. That flexibility is valuable. It also introduces complexity. Halogen takes the opposite approach: ● Target hardware: AMD Strix Halo, using the gfx1151 GPU architecture. ● Target models: The Qwen3.8-Flash-Next model family. ● Optimization strategy: Hardware-specific kernels and model-aware execution paths. In its current implementation, Halogen is deliberately specialized rather than a universal inference solution. Why does that matter? Every layer of generality creates engineering trade-offs. A general-purpose engine must accommodate different hardware capabilities, maintain compatibility paths, and handle alternative execution strategies. A specialized engine can focus its resources on the exact hardware and workload it is designed to run. Think of it as the difference between a Swiss Army knife and a precision-engineered tool. One can do almost everything; the other is built to do one job exceptionally well. 4. Optimization No. 1: Removing Unnecessary Abstraction A general-purpose inference engine such as llama.cpp must accommodate a wide range of GPUs and computing environments. Its implementation needs to remain flexible enough to support different architectures and execution requirements. Halogen can take a more direct route. By developing kernels specifically for the gfx1151 architecture, the engine can optimize critical operations around the platform’s actual hardware characteristics, including its compute resources, memory behavior, and cache hierarchy. This reduces reliance on generic execution paths and allows performance-critical operations to be tuned for the target platform. The advantage isn’t simply about removing a few compatibility checks. It’s about engineering the execution path around the hardware from the ground up. When those optimizations are applied to operations that run repeatedly during inference, even small improvements can compound into a meaningful increase in generation speed. 5. Optimization No. 2: Smarter Execution for Mixture-of-Experts Models Qwen3.8-Flash-Next uses a Mixture-of-Experts (MoE) architecture, with close to 180 billion total parameters. MoE models work differently from conventional dense models. Instead of activating every parameter for every token, they route each input through a selected subset of specialized experts. Imagine a team of specialist doctors. Not every doctor needs to examine every patient. The system identifies the relevant specialists and brings them into the process. This architecture can make very large models practical to run locally—but it also creates specific challenges for inference engines. Expert routing, weight access, and memory movement all affect performance. On a platform where memory bandwidth is a critical resource, unnecessary data movement can become particularly expensive. Halogen is designed around the target model's execution patterns, optimizing expert routing and data movement so that the inference process focuses on the weights and operations needed for the current computation. The goal is straightforward: move less unnecessary data, reduce avoidable overhead, and keep the hardware working efficiently. Fine-Grained Quantization: 5.53 Bits per Weight Halogen also uses a 5.53 bits-per-weight (bpw) quantization configuration. Quantization reduces the amount of memory required to store model weights by representing parameters with fewer bits. This can ease memory pressure and reduce the amount of data that needs to move during inference. But more aggressive quantization isn't automatically better. Pushing precision too low can affect model quality, depending on the model and workload. A 5.53 bpw configuration offers a different balance: compact weight storage without simply forcing the entire model into a uniform 4-bit representation. Combined with model-aware execution, this helps manage the memory demands of large local models while maintaining a practical balance between efficiency and output quality. 6. Optimization No. 3: Speculative Decoding Another important technique is speculative decoding. In conventional autoregressive generation, a language model generates tokens sequentially, with each new token depending on the preceding context. Speculative decoding changes the process. A smaller draft component proposes several candidate tokens in advance. The main model then verifies those candidates in a more parallelized step. Tokens that pass verification can be accepted together, allowing the system to generate multiple tokens for the computational cost of fewer sequential decoding steps. The result depends on how accurately the draft component predicts the main model’s output. When predictions are accepted at a high rate, generation can become substantially faster. For workloads with repetitive or predictable output patterns—such as code completion, structured text generation, and document rewriting—this technique can be especially effective. In GMKtec's internal testing, Halogen reached a peak of 94 tok/s in repetitive-content scenarios. That is a peak result for a specific workload, rather than a universal generation speed. Actual gains depend on the task, context, and draft acceptance rate. Think of it as having an assistant prepare a draft while you review it. When the draft is right, you can approve several words at once instead of composing every word from scratch. 7. The Combined Result: Fast Prompt Processing and Sustained Generation The real benefit comes from combining these optimizations rather than relying on a single trick. On the GMKtec EVO-X5 Pro, equipped with up to 192GB of unified memory, Halogen delivers the following results in GMKtec's internal testing: <| |> ● Around 75 tok/s sustained generation speed ● Up to 2,000 tok/s prompt-processing speed ● Up to 94 tok/s peak generation speed in repetitive-content scenarios These figures describe different stages or conditions of inference and should not be treated as directly interchangeable. Prompt processing, also known as prefill, is the stage in which the model processes the input context. Generation, or decoding, is the stage in which it produces the response token by token. Fast prompt processing helps when you submit a long document, a large codebase, or an extensive conversation history. Fast sustained generation makes the response itself feel more immediate. Together, they help deliver a smoother experience across long-context AI workflows. What Does This Mean for a 128K-Token Context? A 128K-token context can accommodate a substantial amount of material—potentially hundreds of pages, depending on the language, formatting, and content. With prompt processing reaching up to 2,000 tok/s under the tested conditions, the EVO-X5 Pro can ingest large inputs quickly. Its sustained generation performance then helps turn that context into useful answers, summaries, analysis, or code. For developers and professionals who work with large documents or complex AI workflows, both stages matter. A model that generates quickly but takes too long to process the prompt can still feel sluggish. Optimizing the entire inference pipeline makes a more meaningful difference. 8. The EVO-X5 Pro: More Than a Powerful Chip The GMKtec EVO-X5 Pro is built for users who want to run demanding AI workloads locally, with substantial compute resources and memory capacity available on the desktop. Paired with Halogen, it offers more than the hardware alone: a specialized software stack designed to extract additional performance from a specific combination of GPU architecture and model family. This is particularly relevant for developers running local AI agents, coding assistants, and long-context workflows where responsiveness directly affects productivity. Local inference also gives users greater control over where their data is processed. For organizations handling proprietary code, internal documents, or sensitive business information, keeping compatible AI workloads on local hardware can help reduce reliance on cloud-based inference services. The trade-off is deliberate. General-purpose engines prioritize broad compatibility. Halogen prioritizes performance on its target platform and supported models. Neither approach is universally better. But if your workload matches what Halogen is designed to accelerate, specialization can deliver a tangible advantage. Because being able to run a large model and being able to run it smoothly are two very different things. Memory bandwidth is determined by the hardware. How much performance you unlock depends on the software. With the EVO-X5 Pro and Halogen, GMKtec is pushing local AI inference toward a more responsive, desktop-based experience—without requiring a different chip to make the difference. Experience Local AI at Full Speed Discover the GMKtec EVO-X5 Pro and explore what a specialized AI inference engine can bring to local LLM workloads. GMKtec EVO-X5 Pro — Built for the Next Generation of Local AI.
Read morePushing Desktop AI Further: GMKtec EVO-X5 Pro Accelerates 320B GLM-5.3-Flash Inference
A 10× improvement in prefill speed and a 1.85× boost in decoding performance bring large-scale local AI inference to a new level. As AI models continue to scale into the hundreds of billions of parameters, running them efficiently on a desktop system remains a significant challenge. Large models demand substantial memory capacity, high bandwidth, and carefully optimized inference software to deliver a responsive user experience. GMKtec is pushing those boundaries with the EVO-X5 Pro Desktop AI Supercomputer. Through extensive software-level optimization by our engineering team, the EVO-X5 Pro has achieved significant performance improvements when running the 320B-class GLM-5.3-Flash model: ● Approximately 10× faster prefill for processing input prompts. ● Approximately 1.85× faster decoding for generating responses. These results demonstrate how close hardware and software integration can unlock more of the potential of large-scale AI inference on a desktop system. Built for Large Models: 192GB of Unified Memory Running a model at this scale requires more than raw processing power. Memory capacity is a critical factor in enabling large models to run locally, particularly when working with long prompts and extensive context. The GMKtec EVO-X5 Pro is powered by the AMD Ryzen AI Max+ PRO 495 processor, featuring integrated AMD Radeon 8065S graphics, and supports up to 192GB of unified memory. This large, shared memory pool gives demanding AI workloads more room to operate, helping accommodate large model weights, context, and inference workloads within a single desktop platform. By combining high-capacity unified memory with powerful CPU and integrated GPU resources, the EVO-X5 Pro provides a foundation for experimenting with and deploying large language models locally—without relying exclusively on cloud-based AI services. Software Optimization That Makes a Difference Hardware provides the foundation, but inference performance also depends on how efficiently the software uses available resources. To improve GLM-5.3-Flash performance, GMKtec's engineering team carried out targeted, low-level optimizations across the inference stack. The results show substantial gains in two key stages of model inference. Prefill: Approximately 10× Faster Input Processing Prefill is the stage in which a model processes the input prompt before beginning to generate a response. For long documents, large codebases, and context-heavy workflows, this stage can have a significant impact on how long users wait before generation begins. With the optimized configuration, the EVO-X5 Pro achieved an approximately 10× improvement in prefill performance compared with the baseline configuration used in GMKtec's testing. This can substantially reduce input-processing time and help make interactions involving large amounts of context more responsive. Decode: Approximately 1.85× Faster Response Generation Once prefill is complete, the model enters the decoding stage, generating its response token by token. GMKtec's optimizations delivered an approximately 1.85× improvement in decoding performance in the tested configuration. Faster decoding can make a practical difference in extended conversations, long-form writing, code generation, and other workloads that require the model to produce substantial amounts of output. Together, these improvements address two important parts of the inference experience: processing the information you provide and generating the response you need. Stress-Tested with Nearly 500K Tokens of Context Large models must do more than perform well with short prompts. They also need to handle demanding context lengths while maintaining consistent generation performance. To evaluate the optimized GLM-5.3-Flash configuration, GMKtec tested the EVO-X5 Pro across a range of input lengths, from 1,024 tokens to 499,712 tokens—nearly 500K tokens. The results highlight the system's ability to sustain high throughput across a broad range of context sizes. Input length Prefill speed Decode speed Time to first token 1,024 tokens 494.54 tokens/s — 2.22 seconds 499,712 tokens 628.91 tokens/s 28.23 tokens/s 795.77 seconds Source: GMKtec internal testing. Results reflect the tested hardware, software, model configuration, and inference settings. Actual performance may vary. At the shortest tested input length, the system processed the prompt at 494.54 tokens per second, with a time to first token of just 2.22 seconds. At the other end of the test, the EVO-X5 Pro processed an input of 499,712 tokens while sustaining a reported prefill throughput of 628.91 tokens per second and a decoding speed of 28.23 tokens per second. Time to first token was 795.77 seconds. That longer wait reflects the substantial amount of input the model must process before it can begin generating a response. For workloads involving exceptionally large context windows, the ability to process that volume of information locally is an important part of the performance picture. Across the tested context lengths—including 4K, 32K, and 128K tokens—decoding speeds remained in the 28–33 tokens-per-second range, according to GMKtec's internal results. These figures illustrate the importance of evaluating AI systems across different prompt sizes, rather than relying on short-prompt benchmarks alone. A Desktop AI Platform for Private, Local Workflows The EVO-X5 Pro is designed to be more than a conventional mini PC. It serves as a compact desktop AI computing platform for users who want to explore large models, build local AI workflows, and keep more of their processing on their own hardware. Potential applications include: ● Research and document analysis: Work with extensive source material and long-context prompts. ● Software development: Explore AI-assisted coding and code analysis using locally hosted models. ● AI experimentation: Test large language models and develop custom inference workflows. ● Privacy-conscious computing: Keep supported AI workloads and their data on local hardware, reducing dependence on third-party cloud inference services. Local execution can offer greater control over data and deployment, although privacy ultimately depends on the software, network configuration, and services used. For developers, researchers, and AI enthusiasts who want to push beyond conventional desktop AI workloads, the EVO-X5 Pro combines substantial memory capacity with targeted inference optimization in a compact form factor. Bringing Large-Scale AI Closer to the Desktop The progress demonstrated with GLM-5.3-Flash highlights what can be achieved when capable hardware is paired with software engineered to make better use of it. With support for up to 192GB of unified memory, the AMD Ryzen AI Max+ PRO 495 processor, and significant gains in both prefill and decoding performance, the GMKtec EVO-X5 Pro offers a powerful platform for exploring large-scale local AI inference. The next generation of AI computing doesn't have to live exclusively in the cloud. With the EVO-X5 Pro, GMKtec is bringing more of that capability to the desktop. Discover the GMKtec EVO-X5 Pro and explore the possibilities of local AI computing.
Read more





