The page has been translated by Gen AI.

Overview

Service Overview

AIOS provides an environment for developing AI applications using LLM on Virtual Server, GPU Server, and Kubernetes Engine resources created on Samsung Cloud Platform, without the need for separate LLM service installation or configuration.

Key Features

Convenient LLM Usage: Provides LLM Endpoints by default that allow direct use of LLM on Virtual Server, GPU Server, and Kubernetes Engine resources on Samsung Cloud Platform.
Improved AI Development Productivity: AI developers can use various models with the same API, and compatibility with OpenAI and LangChain SDKs allows easy integration with existing development environments and frameworks.
ServiceWatch Integration: Data can be monitored through the ServiceWatch service.

Service Architecture

Provided Features

The following features are provided:

AIOS LLM Endpoint Provision: When applying for Virtual Server, GPU Server, or Kubernetes Engine services, LLM Endpoint information and usage guides are provided on the detail page of the created resources. You can access and use LLM on those resources by following the usage guide.
AIOS Report Provision: You can check the number of calls and token usage by type, resource, and model, as well as total usage by LLM.

Provided Models

The LLM models provided by AIOS are as follows:

Model Name	Model Type	Description	Main Use Cases	Features
gpt-oss-120b	Chat+Reasoning	Latest GPT series open-source model based on 120 billion parameters	Research/experiments, large-scale language understanding, AI services requiring complex reasoning/analysis, building agent-type systems	Ultra-large parameters Broad knowledge coverage, general-purpose application possible Complete CoT chain generation
Qwen3-Coder-30B-A3B-Instruct	Code	Qwen3 series code model optimized for code generation and debugging	Software development, AI code assistant, long document/repository analysis	Large-scale code knowledge learning Multilingual support Long-context understanding possible
Qwen3-30B-A3B-Thinking-2507	Chat+Reasoning	Qwen3 model enhanced for long-form reasoning and deep thinking	Research, analysis reports, logical writing, mathematics, science, coding	Specialized in long-form and complex reasoning Consistent CoT chain generation
Llama-4-Scout	Chat+Vision	Latest Llama model with multimodal capabilities	Document analysis/summarization, customer support/chatbots	Multimodal (text+image), fast inference, single GPU operation possible Ultra-long text, multi-document summarization/analysis possible, multimodal support Top-tier performance in various benchmarks Up to 4 images can be input
Llama-Guard-4-12B	moderation	Key security and moderation model for enhancing reliability and safety in the latest large language models and multimodal AI services	Used for automatic filtering of harmfulness in user input and model responses	Multimodal security classification Specialized in content moderation Multilingual support
bge-m3	embedding	Key embedding model with three characteristics: multi-functionality, multilingual support, and large input capacity	Used when retrieving external knowledge and providing answer evidence in generative AI, combining Dense and Sparse search to ensure both accuracy and generalization performance	Multi-Functionality: Dense embedding retrieval, token-based weighted retrieval (Sparse Retrieval), multi-vector retrieval (Multi-Vector Retrieval) Multi-Linguality: Supports 100+ languages Multi-Granularity: Can process up to 8,192 tokens
bge-reranker-v2-m3	rerank	Key component for various information retrieval, question answering, and chatbot systems that require fast and accurate search result reranking in multilingual environments	Rerank candidate answers or documents for questions in relevance order	Lightweight and fast inference Multilingual support Easy integration: Hugging Face Transformers, FlagEmbedding compatible

Table. AIOS Provided LLM Models

Regional Availability

AIOS can be provided in the following environments:

Region	Availability
Korea West (kr-west1)	Available
Korea East (kr-east1)	Not Available
Korea South1 (kr-south1)	Not Available
Korea South2 (kr-south2)	Not Available
Korea South3 (kr-south3)	Not Available

Table. AIOS Regional Availability

Prerequisite Services

This is a list of services that must be configured in advance before creating this service. For detailed information, please prepare in advance by referring to the guides provided for each service.

Service Category	Service	Detailed Description
Compute	Virtual Server	Virtual server optimized for cloud computing
Compute	GPU Server	Virtual server suitable for tasks requiring fast computation speed such as AI model experimentation, prediction, and inference in cloud environments
Compute	Cloud Functions	FaaS (Function as a Service) based on serverless computing
Container	Kubernetes Engine	Service providing lightweight virtual computing and containers and Kubernetes clusters to manage them

Table. AIOS Prerequisite Services

AI-ML

ServiceWatch Metrics