

WinFast GPU Workstation and Server
Scalable and Optimized System Solution for AI, HPC, AIoT, Big Data and Edge Computing
With market proven GPU and AI expertise, Leadtek provides not only complete solution but also end-to-end certified instructor-led training, workshop, reference architecture and benchmark on NVIDIA GPU-accelerated application to help customer for early adoption.

24-hour online service
Leadtek provides 24/7 online service for customers who purchase system products with pre-installed software. If there are problems related to the use of the product, you are welcome to create a project tracker on the online service system and describe your issue in detail. You can also upload photos as additional explanation. Professional engineers in Leadtek are always available to resolve your problems online.
24-Hour Online Service
Leadtek NVIDIA-Certified System for Enterprise
Systems certified by NVIDIA for accelerated computing
Enterprises today need a scalable computing infrastructure that provides the performance, manageability, security, and scalability to deliver cutting-edge products and services, while increasing their operational efficiencies. NVIDIA-Certified Systems™ enable enterprises to confidently deploy hardware solutions that securely and optimally run their modern accelerated workloads.
NVIDIA-Certified Workstations
Leadtek NVIDIA-Certified desktop workstations are validated for optimum performance and exceptional reliability. They are ideal for running software in areas such as AI, data science, rendering, and visual design.

AI/ML MODEL DEVELOPMENT

ADVANCED VFX

RENDERING

DATA ANALYTICS

ADVANCED VISUALIZATION

PROFESSIONA BROADCASTING
WS750 | WS850 | WS950 | WS1050 | WS2050 | |
---|---|---|---|---|---|
GPU |
|
|
|
|
|
# of GPU | 1 | 2 | 4 | 2 | 4 |
CPU | Intel® i7 14700 (8+12)core/2.1GHz x1 | Intel® Xeon® W5-2445/10 Core/3.1GHz x1 | Intel® Xeon® W-3400 processors | Intel® Xeon® 5415+/8Cores/2.9GHz x2 | Intel® Xeon® 5415+/8Cores/2.9GHz x2 |
RAM | 34GB DDR4-4800 ECC UDIMM x4 | 64GB DDR5-4800 ECC RDIMM x4 | 64GB DDR5-4800 ECC RDIMM x8 | 64GB DDR5-4800 ECC RDIMM x4 | 64GB DDR5-4800 ECC RDIMM x4 |
Storage |
|
|
|
|
|
Power supply | 668W Platinum Level Power Supply | 1200W Redundant Power Supplies, Titanium | 2000W Platinum Level Power Supply | 2000W Platinum Level Power Supply | 2000W(1+1) Titanium Level Redundant Power Supplies |
Learn More | Learn More | Learn More | Learn More | Learn More |
NVIDIA-CERTIFIED SERVERS
NVIDIA-Certified servers create the essential platform for the evolution of enterprise data centers, delivering infrastructure that can handle a diverse range of accelerated workloads. The certification test suite exercises the performance and functionality of a configured server by running a set of software that represents a wide range of real-world applications.
COMPUTE
AI training and inferencing, data analytics, HPC
GENERAL PURPOSE
Visualization, rendering, AI, virtual workstationsHIGH-DENSITY VIRTUALIZATION
Virtual desktops, Virtual WorkstationsENTERPRISE EDGE
Edge inferencing in controlled environments
INDUSTRIAL EDGE
Edge inferencing in industrial or rugged environmentsModel | WinFast GS2040T | WinFast GS2045T | WinFast GS2050T | WinFast GS4840 | WinFast GS4850T | WinFast GS4855ST |
---|---|---|---|---|---|---|
GPU | A6000/A100 | A100 | RTX 6000 Ada | A6000/A100 | RTX 6000 Ada | RTX 6000 Ada |
# of GPU | 4/6 | 4 | 4 | 8 | 8 | 8 |
CPU | Intel® Xeon® Gold 5320, 24 Cores/2.2GHz x2 | AMD EPYC™ 7402, 24 Cores/2.8GHz x 2 | Intel® Xeon® Gold 5418Y, 24 Cores/2.0GHz x2 | Intel® Xeon® Gold 5320, 24 Cores/2.2GHz x2 | Intel® Xeon® Gold 5418Y, 24 Cores/2.0GHz x2 | AMD® Genoa 9254 24Cores/2.4GHz x2 |
RAM | 64GB DDR4-3200 RDIMM x8 | 64GB DDR4-3200 RDIMM x8 | 64GB DDR4-4800 RDIMM x8 | 64GB DDR4-3200 RDIMM x16 | 64GB DDR4-4800 RDIMM x16 | 64GB DDR4-4800 RDIMM x16 |
Storage |
|
|
|
|
|
|
Power supply | 2600W (1+1) Titanium Level Redundant Power Supplies | 2200W (1+1) Platinum Level Redundant Power Supplies | 2600W (1+1) Titanium Level Redundant Power Supplies | 2200W (2+2) Titanium Level Redundant Power Supplies | 2700W (2+2) Titanium Level Redundant Power Supplies | 2000W (2+2) Titanium Level Redundant Power Supplies |
Network Adapter | MELLANOX ConncetX-6Dx x 2 | |||||
Ethernet/Infiniband Switch | SN2100- 100 Gigabit Ethernet Switch with 16 QSFP28 ports Plus cables: Mellanox passive copper for 100 Gigabit Ethernet | |||||
Learn More | Learn More | Learn More | Learn More | Learn More | Learn More |
Equipped with NVIDIA GPU for Large Language Model (LLM) solution
When humans use AI, they need to communicate with computers to find out answers. Large language models (LLM) is a tool that allow computers to understand and use human language. It can not only answer your questions, but also write articles, analyze data, and even help you write programs. This is the benefit of LLM.

The applicable scenarios of LLM as follow:
What hardware should I use for training or inference of large language models (LLMs)?
WinFast workstations and servers both are NVIDIA certified systems. It is a complete product line with powerful NVIDIA GPU and equipped with Leadtek RTX AI Software Pack, which fully integrates NVIDIA Library, Framework, drivers and Docker. Just one click, you can enter the AI development environment. Let you develop AI applications more easily and efficiently.
Maximize resource utilization and reduce operating costs, especially designed for LLM
Struggling with developing large language models (LLMs)? Leadtek's AI professional team has test the popular LLM models such as Llama3.1, Mistral and TAME. Through precise hardware configuration and optimization, Leadtek's NVIDIA certified system can efficiently run LLM models of different sizes with fewer resources, which not only greatly improves AI development efficiency, but also helps you achieve lower total cost of ownership(TCO) .
Why choose Leadtek?
Expert certification:WinFast workstations and servers both are NVIDIA certified systems with reliable quality.
Total solution:Optimized for LLM, the software and hardware are perfectly matched and the performance is excellent.
Rapid deployment:Start AI development environments such as docker with one click, saving time and effort.
Diversified configurations:Provides a variety of system configurations to meet different request.
Professional AI technical team:Leadtek provide professional assistance and suggestions to meet your needs.
Who needs LLM?
AI researchers, AI developers, scientists, engineers, AI departments in enterprise, research institutions in college, etc.
Recommended system configuration
WinFast RTX AI system | Entry-level | Mainstream | High-end |
---|---|---|---|
Support Model | Llama3 8B、Mistral 7B/12B、TAME 8B | Llama3 8B、Mistral 7B/12B、TAME 8B | Llama3 70B (16bit)、Mistral 7B/12B、TAME 70B(16bit) |
Features | Finetune/Inference: Llama3/TAME 8B(16 bit)、Mistral 7B/12B (16bit) | Finetune/Inference | Finetune/Inference: Llama3/TAME 70B (16bit) 、Mistral 7B/12B (16bit) |
Product name | WS850 | WS2050 | GS4850T |
GPU | RTX 6000 Ada x2 | RTX 6000 Ada x4 | RTX 6000 Ada x8 |
CPU | Intel Xeon W series CPUx1 | Intel Xeon 4th/5th CPU x2 | Intel Xeon 4th/5th CPU x2 |
RAM | 256GB | 512GB | 1TB |
Power supply | 1200W Platinum Level Power Supply | 2000W(1+1) Titanium Level Redundant Power Supplies | 2000W (2+2) Titanium Level Redundant Power Supplies |
![]() |
√ | √ | √ |
Learn More | Learn More | Learn More |
NVIDIA-Powered
Data Science and Deep Learning Workstation
for AI Development
WinFast RTX AI Workstation
Leadtek provides WinFast RTX AI Workstation, a complete line of NVIDIA-powered data science workstations featuring Quadro® RTX™ GPUs and leading class data science software. Maximize productivity and lower the cost of your AI development projects. Highly scalable and pre-configured complete system enable time to market, increase efficiency in the organization and to unlock the value from their combined data sets.
Learn More
Delivering Data Science Value

Maximize Productivity
Take advantage of a fully integrated hardware and software stack that is ready out-of-the-box for instant productivity gains.

Faster Model Development and Training
Iterate and move to production clusters faster with high performance Quadro RTX GPUs.

Framework Diagnostic and System Recovery
Simple and intuitive tool for framework diagnosic and system recovery.

Enterprise-Class
The workstation is backed with 3 year warranty for hardware components and the software stack is optimized and tested for world-class support so researcher and developer can rely on
Integrated with GPU-Accelerated Software
NVIDIA-powered data science workstations come with a comprehensive stack of tested and optimized data science software built on NVIDIA CUDA-X AI. A collection of libraries that enable modern computing applications to benefit from NVIDIA’s Quadro RTX GPU accelerated computing platform. This stack features RAPIDS data processing and machine learning libraries, NVIDIA optimized TensorFlow, PyTorch, Caffe and other leading data science software, providing enterprises with accelerated workflows for faster data preparation and model training.
Core Components and Frameworks






WinFast RTX AI Workstations | WS750 | WS850 | WS950 | WS1050 | WS2050 |
---|---|---|---|---|---|
GPU |
|
|
|
|
|
# of GPU | 1 | 2 | 4 | 2 | 4 |
CPU | Intel® i7-14700 (8+12)core/2.1GHz x1 | Intel® Xeon W5-2445/10 Core/3.1GHz x1 | Intel® Xeon W5-3435X/ 16Core/3.1GHz x1 | Intel® Xeon 5415+/8Cores/2.9GHz x2 | Intel® Xeon 5415+/8Cores/2.9GHz x2 |
RAM | 32GB DDR5-4800 ECC UDIMM x4 | 64GB DDR5-4800 ECC RDIMM x4 | 64GB DDR5-4800 ECC RDIMM x8 | 64GB DDR5-4800 ECC RDIMM x4 | 64GB DDR5-4800 ECC RDIMM x8 |
Storage |
|
|
|
|
|
Power Supply | 668W Platinum Level Power Supply | 1200W Platinum Level Power Supply | 2000W Platinum Level Power Supply | 2000W Platinum Level Power Supply | 2000W(1+1) Titanium Level Redundant Power Supplies |
Learn More | Learn More | Learn More | Learn More | Learn More |
AI/Deep Learning System Solution
The GPU-Optimized Servers for DL Workloads to Accelerate Your AI Journey
Implementing the right AI/deep learning solution can be complex. Leadtek provides versatile server, accelerator, DL software and technical consult as end-to-end solution. Our proven, practical approach, validated solutions and partners, GPU-optimized infrastructure and AI software platform – reduce complexity and help you realize the value of your data faster helping you gain a competitive edge.


Scalable Deep Learning
Offering a wide range of customizable options from desktop to data center to meet your budget and project need.

Pre-Configured System
Leadtek AI/DL system solutions are fully turnkey; GPU-accelerated comprehensive software stack of data science software built on NVIDIA CUDA-X AI with RAPIDS, ML libraries and frameworks. Therefore, you avoid the time-consuming of configuration and setup.
Suggested Configurations
WinFast RTX AI Server | WinFast GS2040T | WinFast GS2045T | WinFast GS2050T | WinFast GS4840 | WinFast GS4850T | WinFast GS4855ST |
---|---|---|---|---|---|---|
GPU | A6000/A100 | A100 | RTX 6000 Ada | A6000/A100 | RTX 6000 Ada | RTX 6000 Ada |
# of GPU | 4/6 | 4 | 4 | 8 | 8 | 8 |
CPU | Intel® Xeon® Gold 5320, 24 Cores/2.2GHz x2 | AMD EPYC™ 7402, 24 Cores/2.8GHz x2 | Intel® Xeon® Gold 5418Y, 24 Cores/2.0GHz x2 | Intel® Xeon® Gold 5320, 24 Cores/2.2GHz x2 | Intel® Xeon® Gold 5418Y, 24 Cores/2.0GHz x2 | AMD® Genoa 9254 24Cores/2.4GHz x2 |
RAM | 64GB DDR4-3200 RDIMM x8 | 64GB DDR4-3200 RDIMM x8 | 64GB DDR4-3200 RDIMM x8 | 64GB DDR4-3200 RDIMM x16 | 64GB DDR5-4800 RDIMM x16 | 64GB DDR5-4800 RDIMM x16 |
Storage |
|
|
|
|
|
|
Power Supply | 2600W (1+1) Redundant Power Supplies, Titanium | 2200W (1+1) Redundant Power Supplies, Platinum | 2600W (1+1) Redundant Power Supplies, Titanium | 2200W (2+2) Redundant Power Supplies, Titanium | 2700W (2+2) Redundant Power Supplies, Titanium | 2000W (2+2) Redundant Power Supplies, Titanium |
Learn More | Learn More | Learn More | Learn More | Learn More | Learn More |
How GPU Speeds Up AI &Deep Learning Computing
Visit Leadtek AI
Workstation
Learn more
Server
Learn moreWinFast Workstation
Extreme Performance for Modeling, Animation and AI/Deep Learning
Overview
WinFast Workstation is designed for technical and creative professionals who demand the highest productivity and reliability for the heaviest of workload. Leadtek best-of-class workstations is optimized and configured with NVIDIA GPU for the workload ranging from entry level workload to multi-tasking in complex production pipeline.
Technology
- Highly optimized with NVIDIA GPU for performance and reliability
- Balanced multi-GPU thermal and acoustic design
- Flexible configuration and GPU for versatile workload demands
Application
- AI/Deep Learning Development
- 2D Image Processing
- CAD Design
- Engineering Simulation
- 3D Animation and Modeling
- Rendering
- VR
Key Features

Form Factor
Slim and Mid Tower

Processor
Single or dual Intel® Xeon® Scalable processors up to 205W TDP

GPU
Up to 4 NVIDIA GPUs with NVLink Technology

Networking
Up to dual 10GbT LAN with IPMI 2.0

System Memory
Up to 16 DIMM slot and 4TB DDR4 memory

Storage Drive
Up to 8 drives with 2.5" and 3.5" SAS3/SATA3 options (4 NVMe optional), 1 M.2 NVMe drive

Power Supply
Up to Redundant Titanium Level Power Supplies

Warranty
3-Year hardware services
Workstation Line
1 CPU + 1 GPU

Entry
- PLM
- CAD
- Basic DCC
- Medical Imaging
- Entry DL/ML
1 CPU + 2 GPU

Mainstream
- Mid-size CAD
- Advanced DCC
- Simulation
- Medical Imaging
- DL/ML
1 CPU + 4 GPU

High End
- Complex CAD
- Complex DCC
- Rendering
- 3D Medical Imaging
- Simulation
- Seismic Exploration
- Large Scale DL/ML
2 CPU + 4 GPU

Supreme
- Complex CAD
- Photo-realistic Rendering
- 3D Medical Imaging
- Simulation
- Seismic Exploration
- Large Scale DL/ML
WinFast GPU Server
Optimized and GPU-accelerated for AI/Deep Learning and HPC
Overview
Built on our years of experience in the server industry, Leadtek introduces a comprehensive range of GPU-centric WinFast GS server solutions - designed to deliver, performance, scalability and reliability in the applications such as AI/Deep Learning, High Performance Compute (HPC), virtual graphics, render farm and other GPU-accelerated intensive workloads.
Technology
- Highly Optimized GPU-to-GPU Interconnect Architectures for best efficiency and low latency
- Optimized multi-GPU thermal designs for highest performance and reliability
- Leading NVIDIA® GPU architectures HGX-2 platform with NVLink™ and NVSwitch™
Application
- AI/Deep Learning
- High Performance Computing
- Scientific Research
- Data Analytics
- 3D Animation and Modeling
- 3D Graphics Accelerated Virtualization
- Rendering

Form Factor
1U, 2U, 4U and 10U Rackmount

Processor
Single or dual Intel® Xeon® Scalable processors up to 205W TDP

NVIDIA GPU
Up to 16 GPUs in 10U

Networking
Up to dual 10GbT LAN with optional IB FDR

System Memory
Up to 24 DIMM slot and 6TB DDR4 memory

Storage Drive
Up to 16 NVMe drives with 2.5" and SAS3/SATA3 options, Up to 2 M.2 NVMe drives

Power Supply
Redundant Platinum/Titanium (96%+) Level PSU

Warranty
3-Year hardware services
Server Line
1U / 2 GPU

1U Server
- Virtual GPU (GRID)
- Remote Workstation
- Small Scale Render Farm
2U / 6 GPU

2U Server
- Virtual GPU (GRID)
- Render Farm
- DL/ML
- High Performance
- Computing
- 3D Medical Imaging
4U / 8 GPU

4U Server
- Cloud Computing
- Large Scale DL/ML
- High Performance
- Computing
4U / 8 GPU

4U Server
- Cloud Computing
- Large Scale DL/ML
- Render Farm
- High Performance
- Computing
4U / 10 GPU

4U Server
- Cloud Computing
- Large Scale DL/ML
- High Performance
- Computing
10U / 16 SXM3 GPU

10U Server
- Hyper Scale Cloud
- Hyper Scale DL/ML
- High Performance
- Computing
Leadtek AI Forum
Realizing Your Future of AI
Visit Leadtek AI Forum
What We Do
Leadtek AI Forum is an online discussion community where accessible to Leadtek’s customer or registered developer discussing their queries related to Leadtek’s AI solutions or specific ML/DL technologies to help individual AI development and deployment.
Why Join
Whether you’re looking for ML/DL solutions for your AI project, the Leadtek AI Forum is the place to be. With the entire ecosystem including software and hardware, discuss ML/DL technological evolution, trend, or solution with Leadtek experienced AI experts, NVIDIA DLI certified instructors to advance your AI investment.