Yobitel Multi-Model Text-to-Image Inference Server

The Yobitel Multi-Model Text-to-Image Inference Server is a fully pre-configured, GPU-accelerated AMI designed for AI researchers, artists, and developers who need cutting-edge text-to-image generation and visual synthesis on AWS.This AMI integrates multiple diffusion-based models such as Stable Diffusion 3.5-Medium, FLUX.1-dev, and SDXL-Lightning. These are all pre-loaded and optimised for high-performance GPU inference.

Setting up a multi-model inference environment can be complex due to dependencies like CUDA, PyTorch, and container configurations. The Yobitel AMI simplifies this by providing a production-ready solution so users can start generating images immediately through a browser-based Gradio interface.

Key Features

• Pre-installed AI Models:  Stable Diffusion 3.5-Medium – High-fidelity, realistic text-to-image generation.FLUX.1-dev – Advanced diffusion model enabling creative and design-driven visual outputs.SDXL-Lightning – Optimised for lightweight, high-speed inference and rapid Prototyping
• GPU-Accelerated Inference: Integrated with NVIDIA GPU drivers, CUDA Toolkit, and PyTorch for seamless hardware acceleration and real-time image generation.
• Unified Web Interface: Each model is deployed in a dedicated Docker container and accessible via a Gradio web interface, allowing users to perform inference directly through their browser.
• Customizable & Scalable: Built on an open-source foundation, enabling model fine-tuning, container modification, and scalable deployment across AWS GPU instances.
• Secure & Flexible Configuration: No embedded credentials or SSH keys. One-time password authentication and isolated containerised architecture ensure a secure and controlled environment.
• Comprehensive Documentation: Step-by-step instructions provided for deployment, accessing the web interface, model selection, and customisation.

Technical Usage Manual

1. Once you subscribe to the AMI for Yobitel Multi-Model Text-to-Image Inference Server  Model from the AWS Marketplace, choose the launch through EC2 and launch.
2. It redirects to the launch instance page, configures the required details, i.e., Name, Instance type, Keypair, Network Setting, Storage, and launches the Instance.
3. Choose the Instance type as per your requirements.
4. When the instance is successfully created, go to the EC2 Dashboard in the AWS  Console, select your created instance, and copy the public IP of the instance.
5. After launching the instance, place the public IP in your terminal .
6. It Displays Three models in the terminal, you can choose one that you want.
7. You can choose the model that you want by entering the option number.
8. It takes some time to launch the model. Once it launches, it will show the port .
9. Before placing the public IP in the browser, make sure that your AWS instance security port is open for 7861,7862, and 7863 ports.
10. Use the obtained public IP address from the created instance (Ex, 38.84.57.81).
11. For Each Model, we have different ports like          (http://:7862) for Stable  Diffusion 3.5  (Model1),            (http://:7861)  for FLUX.1 (Model 2 ) ,           (http://:7863) for  SDXL Model (Model 3)
12. Open a web browser. Navigate to: (http://:7862). Replace your instance's public IP in  ().
13. Now the Selected Model will be launched in your browser.
14. If you prefer, you can also preload your desired model using our configuration file and intuitive web interface.
15. The user can launch one model at a time.

Insights & Support

• We will do our best to respond to your questions within the next 24 hours in business days. For any technical support or queries, please contact our Support.
• Please check out our other Containerised Cloud-Native application stacks, such as Blender 3D GPU rendering, Paraview 3D Solutions, GPT-OSS-20B  and AMI - Amazon Machine Images- in the AWS Marketplace.