PAI Model Gallery lets you deploy, fine-tune, distill, and build applications with DeepSeek models — without managing infrastructure. It provides one-click deployment and fine-tuning for large language model (LLM) variants across different resource tiers, from free trial GPU instances to high-performance Lingjun clusters.
Available DeepSeek models
Choose the model that fits your use case and available resources.
|
Model |
Description |
Resource requirements |
|
DeepSeek-V3 |
Full-size DeepSeek V3 model |
Higher compute resources required |
|
DeepSeek-R1 |
Full-size DeepSeek R1 reasoning model |
Higher compute resources required |
|
DeepSeek-R1-Distill-Qwen-7B |
Distilled model based on Qwen2.5-7B. Recommended for quick trials |
Low. Can be deployed with free trial resources |
Before you start
Activate PAI and create a workspace
PAI workspaces centralize management of computing resources, permissions, and AI assets. Activating PAI creates a default workspace. Object Storage Service (OSS) is activated by default for storing code, models, and datasets.
-
Go to the PAI console to activate PAI. Each region requires separate activation.
-
(Optional) If OSS is not activated, go to the OSS console and create a bucket in the same region as your PAI workspace.
-
If no workspace exists, go to the PAI Workspaces page and click Create Workspace.
-
Go to Model Gallery to find your model.
Region and resource specifications
PAI workspaces and OSS buckets are region-specific, and some regions are not interconnected. Select your region carefully.
Key considerations:
-
Resource availability varies by region. If resources are unavailable in one region, check other regions.
Billing: PAI offers pay-as-you-go and subscription billing. Pay-as-you-go resources are shared, so shortages may occur.
Restricted specifications: Some resource specifications are restricted to whitelisted users. Contact your sales manager for recommendations.
Lingjun resources: PAI also supports Lingjun AI Computing Service resources with high-speed networks for distributed training or deployment. Lingjun resources are restricted to whitelisted users. Contact your sales manager if needed.
(Optional) Create a virtual private cloud (VPC) for distributed training or deployment
A VPC is required for distributed training or deployment. For internet access, configure a public port.
When creating a VPC, also configure a vSwitch and security group:
Specify your VPC when starting training or deployment in Model Gallery. 
Deploy a DeepSeek model
One-click deployment is available for DeepSeek-V3 and DeepSeek-R1 models. For detailed instructions, see One-click deployment of DeepSeek-V3 and DeepSeek-R1 models.
To get started quickly, try DeepSeek-R1-Distill-Qwen-7B. This distilled model has low resource requirements and can be deployed with free trial resources.
Fine-tune and distill a DeepSeek model
Fine-tuning trains the model on your data to improve accuracy for a specific use case.
Distillation transfers knowledge from a larger teacher model to a smaller student model, retaining accuracy while reducing compute and storage costs.
Fine-tuning success depends on dataset quality, hyperparameters, and experimentation. For many use cases, retrieval-augmented generation (RAG) may be simpler and sufficient.
For detailed instructions, see One-click fine-tuning of DeepSeek-R1 distill models.
Build AI applications with LangStudio
Develop applications with LangStudio
PAI LangStudio simplifies enterprise LLM application development with built-in templates for RAG, web search, and other application types.
The following tutorials cover common application patterns using DeepSeek models in LangStudio:
|
Application pattern |
Tutorial |
|
DeepSeek + Knowledge base |
Use LangStudio to create a DeepSeek- and RAG-based Q&A application flow for finance and healthcare |
|
DeepSeek + Web Search |
|
|
DeepSeek + Knowledge base + Web Search |
Use LangStudio and DeepSeek to deploy a RAG- and web search-based chatbot |