Enterprise Sovereign AI Inference Cloud Platform
Flexible Leasing • Secure Isolation • Diverse Models • Rapid Deployment
A unified platform integrating compute, models, and deployment environments. We empower enterprises to adopt AI rapidly—perfectly balancing performance, security, and flexibility.
Core Features: Three Key Advantages
Feature 1
Isolated Private Virtual Environments
Security equivalent to on-premises environments, combined with the flexibility and convenience of the cloud.
Physical-Level Isolation: Each client receives an independent, private Virtual Machine (VM) environment. Data and computing resources are completely isolated from others, ensuring zero compromise on security.
Local Compliance: Our data centers are located entirely within Taiwan, fully complying with strict data residency regulations required by government, finance, healthcare, and other highly regulated industries.
Multi-Layered Protection: Integrates cloud firewalls, OS isolation, and comprehensive audit logging to achieve the highest level of privacy protection.
Feature 2
Hourly Billing for Maximum Cost Optimization
Eliminate token-based billing anxiety and gain precise control over your AI budget.
Low-Barrier Flexible Leasing: Withour "pay-by-the-hour" model, there are no heavy upfront hardware costs. You can turn off your instances at any time when they are not inuse.
Unlimited Inference: Enjoy dedicated computing power during your leasing period with unlimited token inference. Never worry about skyrocketing costs as your usage scales.
Transparent Analytics: Built-in token usage dashboards allow you to track consumption and costs byAPI key.
Feature 3
One-Click LLM Installation & Management
Deploy AI applications in minutes—no MLOps expertise required.
Intuitive GUI: An intuitive graphical user interface lets you download, activate, and switch models with just a few simple clicks.
Ultra-Fast Image Booting: Leverage exclusive AFS AI Hub images to complete automated system configuration in minutes.
OpenAI API Compatibility: Seamlessly migrate existing applications to your private cloud environment with minimal transition costs. Supports management via multiple API keys.
Powerful Compute Infrastructure & Model Library
Enterprise-Grade AI Compute: On-Demand, Scalable, and Instant
Spin up GPU compute instantly based on your needs, eliminating high upfront hardware investments and complex deployments. We provide high-performance, cost-effective AI computing resources optimized for Large Language Model (LLM) inference, AI Agents, and Generative AI applications.
Support for Diverse Mainstream Models
Our platform continuously updates its model registry, covering key categories including Language, Vision, Reasoning, and Vector/Reranking:
Language (LLM)
Llama 3.3-FFM-70B
Llama 3.1-FFM-8B
Vision (VLM):
Llama 3.2-FFM-11B
Gemma 3-4B
Gemma 4 26B
Gemma 4 31B
Reasoning
gpt-oss 120B
gpt-oss 20B
Vector & Sorting Model
Embedding-v2.1
llama-nemotron-embed-vl-1b-v2
llama-nemotron-embed-vl-8b
llama-nemotron-rerank-1b-v2
Ready to Elevate Your Business with AI?
Start Your AI Journey Quickly in 4 Simple Steps
STEP 1
Create Your Account
Register on the Taiwan AI Cloud website to createU-T Cloud Accountand purchase your compute credits.
STEP 2
Spin Up Your VM
Select the AFS AI Hub image and your preferred NVIDIA H100 GPU specifications.
STEP 3
Deploy & Run
Connect to your instance and run the startup command; the system will configure itself automatically.
STEP 4
Start Inference
Open the intuitive browser-based GUI, download your models, and launch your AI services instantly.
FAQ
How does AFS AI Hub Cloud ensure the security of sensitive enterprise data?
We provide each client with an independent, private virtual environment, ensuring your data and computations are completely isolated from others. Additionally, our data centers are located within Taiwan, strictly complying with local regulations regarding data residency for government and specific highly regulated industries.
Will my inference data be recorded or used for model training?
Absolutely not. Our system features a multi-layered security architecture, including cloud firewalls and OS-level isolation. Furthermore, comprehensive logging and auditing capabilities give your enterprise full control over all activity records, ensuring your trade secrets remain completely secure and confidential.
What is the billing model?
AFS AI Hub Cloud Platform operates on an hourly billing model, where you rent dedicated H100 computing resources. During your leased hours, you can run unlimited inferences (token generation) without the constraints of pay-per-use APIs, significantly optimizing your long-term operational costs.
Can I turn off the service at any time when it is not in use?
Yes, absolutely. You can flexibly scale your GPU specifications based on your actual needs and temporarily pause or shut down your instances when they are not in use to save budget and achieve maximum cost optimization.
Do I need advanced AI technical expertise to use the platform?
Not at all. We provide an intuitive graphical user interface (GUI) that allows you to download and activate models with a single click. You can launch your AI applications instantly—no professional MLOps background required.
Which mainstream models are currently supported?
Our platform supports a wide range of state-of-the-art open-source models, categorized to meet diverse business needs across text, vision, and vector processing: Llama 3.3-FFM-70B, Gemma 4 series, and various Vision-Language (VLM) and reasoning engines.
Can I switch to AFS AI Hub if I am currently using the OpenAI API?
Yes, absolutely. The AFS AI Hub API is fully compatible with OpenAI's API format. You can seamlessly migrate your existing applications simply by updating the API Key and Base URL, ensuring near-zero transition costs.