Quick Deployment

Deploy a cloud model by selecting an authorized account, model, inference engine, strategy, and compute plan.

Target Outcome

A deployment request is created with an understood resource plan and cost estimate, then appears in My Deployments.

Applicable Roles

• End User

Before You Start

• Confirm that an authorized access account, region, and deployable model are available.
• Decide whether the workload needs a single-node or high-availability plan and define a cost limit.

Procedure

Step 1: Select a Cloud Account and Model

1. From the platform home page, select Quick Deployment in the left navigation to open the model catalog.
2. In Deployable Scope, select the target cloud and region, such as Alibaba Cloud - China East 2 (Shanghai). The platform filters models for that scope.
3. In Matching Models, use the model, series, and scenario selects, the search box, and the sort control to find the target model.
4. Locate the model row, such as Qwen3-8b, and select Deploy Model to open Step 2.

Parameter Reference

Field	Type	Example	Description
Current Scope	Label	Alibaba Cloud / China East 2 (Shanghai)	Displays the selected cloud-account and region combination
Cloud Filter	Select	All Clouds	Required; filters deployable scopes by cloud provider
Region Filter	Select	All Regions	Required; filters deployable scopes by region
Cloud Provider	Single-select card	Alibaba Cloud - China East 2 (Shanghai)	Required; selects the target cloud account and region
Model Filter	Select	All Models	Optional; filters by model name
Series Filter	Select	All Series	Optional; filters by model series
Scenario Filter	Select	All Scenarios	Optional; filters by application scenario
Search	Text	Qwen3	Optional; locates a model by keyword
Sort	Select	Default Sort	Optional; changes model-list ordering
Model	Model row	Qwen3-8b	Required; select Deploy Model to continue

Step 2: Review the Recommended Deployment

1. After selecting Deploy Model, review the Recommended Deployment page.
2. Confirm the selected-model card, including model name, series, capabilities, version, and context. Select Single Node or High Availability for the deployment mode.
3. Select an inference engine:
   • vLLM for throughput, ecosystem maturity, and OpenAI API compatibility.
   • SGLang for complex inference chains, multi-turn orchestration, and cache-hit optimization.
4. Select a business strategy:
   • Cost Effective prioritizes hourly price and resource utilization.
   • High Performance prioritizes high-performance GPUs, bandwidth, and newer engines.
   • Available Now prioritizes capacity that can start immediately.
   • GPU Count prioritizes multi-card parallelism and scaling requirements.
5. Review the recommended model-compute plan, including provider and region, engine, instance type, GPU, CPU and memory, capacity, and hourly, daily, monthly, and annual estimates.
6. After verifying the engine, strategy, resource plan, and estimated cost, select Confirm Deployment. Track the result in My Deployments.

Parameter Reference

Field	Type	Example	Description
Deployment Mode	Single-select tab	Single Node / High Availability	Required; selects the deployment architecture
Inference Engine	Single-select card	vLLM	Required; selects the model inference framework
Business Strategy	Single-select card	Cost Effective	Required; selects the compute filtering and ranking policy
Model-Compute Plan	Recommendation card	Alibaba Cloud / cn-shanghai / vLLM	Required; calculated from account, region, engine, and strategy
GPU	Label	NVIDIA T4 x 1	Required; GPU model and count
CPU / Memory	Label	4C / 15G	Required; CPU and memory allocation
Hourly Estimate	Number	CNY 12.78/hour	Required; estimated hourly cost
Daily Estimate	Number	CNY 306.72/day	Required; estimated daily cost
Monthly Estimate	Number	CNY 9201.60/month	Required; estimated monthly cost
Annual Estimate	Number	CNY 111952.80/year	Required; estimated annual cost

Completion Checklist

Purpose: These are the exit criteria for the current feature task. Use them to decide whether the result is observable and reviewable and whether you can continue to the next step in the scenario. They do not repeat the procedure; if any item fails, follow the troubleshooting section below.

Check	Pass Criteria
1	Account, region, model, engine, and compute plan are selectable.
2	The request is submitted with the expected cost and scope.
3	My Deployments contains the new record.

Troubleshooting

Symptom	Check First
No model or compute plan can be selected	Account authorization, region, asset compatibility, quota, and current capacity
Deployment creation fails	Required fields, cost selection, cloud-account state, and event details

User Manual

Review the complete Quick Deployment steps, validation rules, and common issues