OpenClaw is an open-source, local-first AI agent and automation platform. OpenClaw is deployed on your own Simple Application Server rather than being provided as a cloud-hosted (SaaS) service. After you purchase a Simple Application Server and select the OpenClaw application image, you can run OpenClaw on your own server. The OpenClaw deployed with one click on Alibaba Cloud Simple Application Server is the open-source community standard version (MIT license), which is identical to the GitHub community version and is not a modified version by Alibaba Cloud. You can configure and extend it freely in the same way as the community. You can use OpenClaw to deeply integrate multi-channel communication capabilities with large language models (LLMs) to create custom AI assistants with persistent memory and proactive execution capabilities.
OpenClaw is an open-source AI assistant. Before you use it, fully assess its security and stability and strictly follow the license agreement to protect your system and data.
Billing
Deploying and using the OpenClaw service on a Simple Application Server instance incurs the following fees:
-
Simple Application Server fees: For more information, see Simple Application Server billing overview.
-
Model invocation fees: These fees are determined by the model provider and the specific model you choose.
-
Connection token: The connection token is automatically generated after the API key is configured successfully. You do not need to purchase any plan.
Prerequisites
The OpenClaw application image supports only Linux (Alibaba Cloud Linux) and does not support Windows.
-
If you already have a Windows Simple Application Server, you can reset the system to the Linux OpenClaw image (select the OpenClaw image under Application Images during the reset). Note: In the China (Hong Kong) and overseas regions, switching between Windows and Linux is not supported. For more information, see Reset a system.
-
If you want to keep Windows, see the official OpenClaw documentation for other deployment methods.
Purchase an OpenClaw package
If you already have a Simple Application Server instance, you can skip this step and deploy OpenClaw by resetting the system. When resetting the system, select the OpenClaw image under Application Images. A system reset erases all data on the system disk, so you must back up your data first. For instructions, see Reset a system. To preserve workspace data such as the AI assistant's memory, identity configuration, and Skills, refer to Back up and restore data for OpenClaw version upgrades. The OpenClaw image has the OpenClaw service pre-installed. Do not install OpenClaw again, because a duplicate installation may cause process conflicts. If a conflict occurs, reset the system to the OpenClaw image.
-
Purchase a Simple Application Server instance with the OpenClaw application pre-installed. OpenClaw, MoltBot, and Clawdbot refer to the same product.
-
On the Simple Application Server page, the resources are preconfigured. You can modify these settings as follows:
-
Servers: The memory must be at least 2 GiB. You can switch packages based on your needs. 2 GiB of memory is sufficient for basic testing and single-skill scenarios. To run multiple skills at the same time or handle high-concurrency access, upgrade to 4 GiB or higher for stable operation.
-
Region: The default is US (Virginia).
-
Purchase Settings: The default subscription period is 12 months. You can change the duration as needed.
-
Selection and deployment recommendations
Before you purchase a Simple Application Server with the OpenClaw image, review the following recommendations to ensure the server meets your requirements.
Operating system support
The OpenClaw application image provided by Alibaba Cloud runs exclusively on Linux (Alibaba Cloud Linux) and does not support Windows. To run OpenClaw on Windows, select a standard OS image and follow the official OpenClaw documentation for manual installation. Alibaba Cloud does not test or provide technical support for OpenClaw on Windows. You may encounter issues with path handling, dependency installation, or service management.
Region selection
The region you select affects the network latency for model API calls. Choose a Simple Application Server region based on the location of your target model service:
Primarily calling China-based models (such as Model Studio or Volcengine): Select a mainland China region to avoid cross-border latency.
Primarily calling models hosted outside mainland China (such as OpenAI) or needing access to overseas websites: Select China (Hong Kong) or an overseas region.
Instance specification recommendations
The default specification link pre-configures a 2-core 2 GB setup (swas.s.c2m2s40b1.linux). This specification is suitable for running a single Agent only. Running multiple Agents or executing complex tasks on a 2-core 2 GB instance may result in insufficient resources and service instability. Upgrade your instance specification based on actual needs.
Multi-instance and multi-Agent support
A single Simple Application Server supports deploying multiple OpenClaw instances or configuring multiple Agents. However, this requires manual configuration file modifications or following third-party tutorials. The pre-installed image does not provide a web-based multi-tenant or multi-Agent management interface.
Configure OpenClaw
The Application Details tab is displayed only when the instance uses an application image. If you can see only the Remote Connection button but not the Application Details tab on the instance details page, the instance uses an OS image. Reset the system and select the OpenClaw application image: On the instance details page, in the Basic Information area, click Reset System and select the OpenClaw application image. A system reset erases all data on the disks. Back up your data first.
OpenClaw offers two configuration methods. Select the tab that corresponds to your image version to view the steps.
OpenClaw 2026.5.19 and later (Recommended)
Deploying OpenClaw must be performed in a PC browser. The console is not adapted for mobile devices. You do not need to download standalone console software. You can complete the deployment by visiting the Alibaba Cloud website and logging on in a browser.
-
Log on to the Simple Application Server console. On the server list page, find the target server card and click the instance ID on the card (a 32-character string, such as
2ab6b6fdcbf742ecb4403b83f5f582a0) to go to the Server Overview page. Then, click the Application Details tab.If you cannot find the target server in the server list, check whether the region in the upper-left corner of the page is correct. Servers in different regions are displayed separately. You must switch to the region where the server is located to see the instance.
-
In the lower-left corner of the application details page, click the Initialize button, and complete Step 1: Model Configuration in the initialization wizard that appears.
-
Large Model Platform: Select a model provider from the drop-down list. Supported providers include Alibaba Cloud Model Studio Token Plan (Team Edition), AI coding plan, Alibaba Cloud Model Studio, deepseek, Kimi, GLM Coding Plan, and Zhipu.
-
Before configuring the API Key, create an API key in the Model Studio console: Log on to the Alibaba Cloud Model Studio platform. In the upper-right corner of the page, switch to the target region. Then, go to the Key Management page, click Create API Key, and follow the on-screen instructions to complete the creation. The created API key starts with
sk-. Copy it for the API Key step below. -
API key: Select the region for your API key and enter the API key.
-
Model: Select or manually enter the model name.
After completing the configuration, click Configure Now to proceed.
-
-
In the wizard, under Step 2: Use Web UI, first click Allow in the Allow OpenClaw Access Port area to open the port. Then, click Log on to Web UI to go to the OpenClaw WebUI and interact with the OpenClaw agent. After the port is allowed, you do not need to configure a domain name. You can access the WebUI directly by using the public IP address of the instance and the port.
ImportantAllowing port access exposes the service to the public internet. After initialization, you can disable public access at any time by using the Public Access toggle switch in the status bar at the top of the application details page. For instructions, refer to How do I enable or disable public access for the OpenClaw WebUI?.
-
(Optional) View the token. After initialization, you can copy the full token by clicking the copy icon next to the Token field in the status bar at the top of the application details page.
WarningDo not share the full URL that contains the token. This link includes authentication credentials. Anyone with this link can bypass login verification and gain administrator permissions for your OpenClaw console.
Versions earlier than OpenClaw 2026.5.19
-
Log on to the Simple Application Server console. On the server list page, find the target server card and click the instance ID on the card (a 32-character string) to go to the Server Overview page. Then, click the Application Details tab.
If you cannot find the target server in the server list, check whether the region in the upper-left corner of the page is correct. Servers in different regions are displayed separately. You must switch to the region where the server is located to see the instance.
-
In the OpenClaw Usage Steps section, under Allow Port Access, click Run Command to open the firewall port for the OpenClaw service.
Important-
To prevent malicious scanning and targeted attacks, OpenClaw no longer uses a fixed default port during initialization. Instead, it generates a random port. You can view the OpenClaw port number in the console.
-
Allowing port access exposes the service to the public internet. You can disable public access to the WebUI with a single click. For instructions, refer to How do I enable or disable public access to the OpenClaw WebUI?
-
-
Under Configure OpenClaw, click Run Command to configure the Alibaba Cloud Model Studio API key.
You can configure two types of Alibaba Cloud Model Studio API keys:
-
AI coding plan exclusive API key (Recommended): This plan provides a monthly request quota for a fixed fee. Calls that exceed the quota are blocked, incurring no additional charges and preventing unexpected costs.
The AI coding plan currently supports models such as
qwen3.5-plus,kimi-k2.5,MiniMax-M2.5, andglm-5. For a complete list of supported models, refer to AI coding plan overview. -
Pay-as-you-go Alibaba Cloud Model Studio API key.
Configure the API key. Select an API key for an AI coding plan from the drop-down list or enter a pay-as-you-go API key. After selecting a model, click Next.
If you manually enter an API key, ensure that the selected region matches the region of the API key. Otherwise, model invocations will fail.
ImportantYou cannot configure an Alibaba Cloud Model Studio Token Plan during initialization. Attempting to do so will cause an error. To use a Token Plan, complete the initialization first and then refer to How do I integrate an Alibaba Cloud Model Studio Token Plan (Team Edition) with OpenClaw?.
-
-
Under Access Web UI Panel, click Run Command to retrieve the URL for the OpenClaw WebUI. You can use this URL to interact with the agent in a web browser.
If needed, you can disable public access to the WebUI with one click. For instructions, refer to Disable public access to the OpenClaw WebUI.
-
(Optional) View the token.
Go to and click View to retrieve and save the token.
WarningDo not share the full URL that contains the token. This link includes authentication credentials. Anyone with this link can bypass login verification and gain administrator permissions for your OpenClaw console.
Configure Token Plan team edition
You can integrate Token Plan Team Edition with OpenClaw to manage API access. The configuration method depends on the OpenClaw image version you are running.
Version 2026.5.19 and later
After initialization, you can integrate Token Plan Team Edition through the Web UI model configuration interface or by modifying the configuration file.
Versions before 2026.5.19
The initialization wizard in versions before 2026.5.19 does not support direct configuration of the Token Plan API Key. Attempting to enter it during initialization will cause errors or no response. Complete the basic initialization first, then configure Token Plan by editing the configuration file (/home/admin/.openclaw/openclaw.json). You need to provide the following values:
Base URL: The endpoint for the model API service, for example,
https://dashscope.aliyuncs.com/compatible-mode/v1.API Key: A dedicated Token Plan API Key that starts with
sk-sp.
Key notes
The Token Plan API Key is displayed only once upon creation or reset. Save it securely.
The Base URL region must match the region where the API Key was created. For example, a Beijing region Key corresponds to the Beijing region Base URL. Mismatched regions result in HTTP 401 errors.
After modifying the configuration, connect to your server via SSH and run
openclaw gateway restartfor changes to take effect.
Configure messaging channels
OpenClaw can communicate with you through various chat applications. All channels support text messages, but support for replies with images, videos, and other media types varies by channel.
Earlier OpenClaw image versions (such as 1.27) do not include the channel configuration feature, and the console does not show the integration entries for the messaging channels listed above. If your image version is earlier and lacks the required features, use Reset a system to replace it with the latest OpenClaw image. A system reset erases all data on the system disk. Back up your data first. To preserve the workspace data of the AI assistant, see Back up and restore data for OpenClaw version upgrades.
Personalize OpenClaw
Generic agents provide a standardized experience that is often unsuitable for specific interaction scenarios. To truly integrate your OpenClaw agent into your business workflows, such as for immersive role-playing or formal enterprise customer service, OpenClaw provides personalization capabilities. Alibaba Cloud provides configuration examples for some scenarios. For more information, refer to OpenClaw personalization templates and scenario examples.
The configuration files include:
-
Personality definition (SOUL.md): Defines the AI agent's core values, conversational style (such as formal, humorous, or concise), and behavioral guidelines.
-
Identity definition (IDENTITY.md): Defines the AI agent's name, self-perception, background story, and role.
-
Workflow definition (AGENTS.md): Defines the AI agent's logic for processing tasks, rules for using tools, and workflows.
Procedure: Go to and click Edit to modify the configuration files. For examples of scenario-based configurations, refer to OpenClaw personalization templates and scenario examples.
Version upgrade and image restrictions
Direct upgrade not supported
The OpenClaw application images pre-installed by Alibaba Cloud do not support direct upgrade of the OpenClaw core version through the in-app Update now button or manual commands. Forcing an upgrade disconnects the integration with the Alibaba Cloud console (such as one-click configuration, Token Plan support, and AppFlow) and voids official support.
Upgrade method
To use a newer version, wait for Alibaba Cloud to release a new application image, then select the latest version through Reset System in the console.
Resetting the system clears all data on the system disk. Create a snapshot backup before performing the following steps. For more information, see Backup and Recovery for Upgrading OpenClaw Versions.
Log on to the Simple Application Server console, select your server, and choose Reset System from the More menu.
Select Replace with Another Image and choose the latest version of the OpenClaw application image.
Click OK to confirm the reset. After the reset completes, the server status changes to Running and the application image version shown in the console reflects the new version.
Manual deployment differences
If you choose a clean OS image to manually deploy OpenClaw, the above restrictions do not apply. However, you are responsible for maintaining all configurations and integrations.
FAQ
How do I restart the OpenClaw gateway?
If you encounter a connection interruption or service outage, you can restart the OpenClaw gateway in the console. The procedure varies depending on your image version:
-
For OpenClaw 2026.5.19 and later: Go to the application details page. In the Gateway Status area of the top status bar, click Restart.
-
For versions earlier than OpenClaw 2026.5.19: Go to and click Run Command.
How to view the OpenClaw port?
To prevent malicious scanning and targeted attacks, OpenClaw no longer uses a fixed default port during initialization. Instead, it generates a random port. The procedure to view the port varies depending on your image version:
-
For OpenClaw 2026.5.19 and later: Go to the application details page. You can find the port number directly in the Port field in the top status bar. Click the copy icon to copy the port number.
-
For versions earlier than OpenClaw 2026.5.19: Go to and click Run Command to view the OpenClaw port number.
How to change the OpenClaw model?
You can switch between different models for OpenClaw in the console. The procedure varies depending on your image version:
-
For OpenClaw 2026.5.19 and later: Go to the application details page. In the Models card, find the configured model and click Switch Model to select a different one, or click Add Model to connect to a new model provider. This version supports multiple providers, including AI coding plan, Alibaba Cloud Model Studio Token Plan, Alibaba Cloud Model Studio, deepseek, Kimi, GLM Coding Plan, and Zhipu.
-
For versions earlier than OpenClaw 2026.5.19: Go to , delete the default model, and then select a different Alibaba Cloud Model Studio model from the drop-down list.
How to enable or disable public access?
You can enable or disable public access to the OpenClaw WebUI with a single click as needed. The procedure varies depending on your image version:
-
For OpenClaw 2026.5.19 and later: Go to the application details page and use the toggle switch in the Public Access area of the top status bar.
-
For versions earlier than OpenClaw 2026.5.19: Go to and click Enable / Disable to control public access for the OpenClaw WebUI.
What happens to OpenClaw on server shutdown?
Yes, OpenClaw stops when the server is shut down and automatically resumes when the server is started. If you need to temporarily stop public access to the service without shutting down the server, you can turn off public access for the WebUI by using the Public Access switch on the application details page. For instructions, refer to How do I enable or disable public access for the OpenClaw WebUI?.
Will enabling the memory feature trigger the heartbeat feature?
The memory feature and the heartbeat feature in OpenClaw are completely independent modules. Enabling the memory feature only saves conversation history to give the AI assistant cross-session memory capabilities. It does not trigger or affect the state of the heartbeat feature. You can safely enable the memory feature without worrying about reactivating a disabled heartbeat.
Why is my API key not recognized?
The system automatically detects the Alibaba Cloud Model Studio endpoint that is geographically closest to your server's region and displays the corresponding API key. If you have subscribed to an AI coding plan, its API key is also displayed. If an API key for the nearest endpoint does not exist, follow the on-screen instructions to create an API key in the corresponding region or subscribe to an AI coding plan.
|
Simple application server region |
Recommended API key region |
Coding plan region |
|
Chinese mainland regions |
China (Beijing) |
Singapore |
|
US and European regions |
US (Virginia) |
|
|
Hong Kong (China) and other Asian regions |
Singapore |
Why can't I find the menus or features described in the documentation in the console?
This documentation is written based on the latest OpenClaw image. Earlier image versions (such as 1.27) have fewer features, and some menus (such as channel configuration and the initialization wizard) do not exist in earlier versions. You can view the current image version number at the bottom of the page.
If the version is too early, use Reset a system to replace it with the latest OpenClaw image. After the reset, you can follow the documentation normally. A system reset erases all data on the system disk. Back up your data first.
What do I do if the console interface does not match the screenshots in the documentation?
The Simple Application Server console is isolated by region. The instance list displays only the resources in the currently selected region. If you have purchased a server but cannot see the instance list, check the region selector in the upper-left corner of the console. After you switch to the region where the instance resides, the instance list is loaded automatically. The Buy Now banner at the top of the console is a permanent promotional banner and is displayed regardless of whether you have instances. It does not indicate that you have not purchased a server.
Can I deploy OpenClaw on my own server without using the application image?
Yes. OpenClaw is an open-source project. You can deploy it on your own server without using the Alibaba Cloud application image. For deployment instructions and configuration guidance, see OpenClaw documentation on Model Studio.
Self-deployed OpenClaw does not include the managed features provided by the Simple Application Server console, such as the initialization wizard, channel configuration, and model management. You are responsible for environment configuration and ongoing operations.
How do I deploy OpenClaw on a Windows system?
The OpenClaw application image provided by Alibaba Cloud runs exclusively on Linux (Alibaba Cloud Linux) and does not support Windows. If you want to use OpenClaw on a Windows system, see the official OpenClaw documentation for alternative deployment methods.
What is the difference between deploying OpenClaw on Simple Application Server and on WUYING Workspace?
Simple Application Server runs OpenClaw in server mode: the instance has a public IP address, and you can access the OpenClaw WebUI directly in a browser by using the public IP address of the server. It is suitable for scenarios that require later configuration changes or externally provided services. WUYING Workspace runs OpenClaw in local mode: similar to an ordinary local computer, it has no fixed public IP address, and you must log on to the cloud desktop and open the OpenClaw application to use it.
For simple usage, WUYING Workspace is easier to operate. If you need to modify configurations later or provide external access, we recommend that you choose Simple Application Server (server mode). A 2-core 2 GB configuration is sufficient for common scenarios such as information search, customer service, and material search.
What do I do if I cannot find the OpenClaw features?
The configuration entries and running status of OpenClaw are located on the Application Details tab, not on the Server Overview page. It is normal that you cannot see OpenClaw-related features on the Server Overview page. This does not mean that the deployment failed.
Locating path: Log on to the Simple Application Server console. On the server list page, click the instance ID on the target server card to go to the Server Overview page. Then, click the Application Details tab to view the configuration entries and status of OpenClaw.
If you can see only the Remote Connection button but not the Application Details tab on the instance details page, the instance uses an OS image. You must reset the system and select the OpenClaw application image first.
If the Application Details tab exists but specific menus described in the documentation (such as channel configuration and the initialization wizard) are missing, see Why can't I find the menus or features described in the documentation in the console?. If the server list itself is empty, see What do I do if the console interface does not match the screenshots in the documentation?.
What do I do if the token is empty?
The token is displayed in the Token field in the status bar at the top of the Application Details page. It is automatically generated after the API Key is configured successfully. You do not need to purchase any plan. Click the copy button to copy the full token.
If the Token field is empty, check the following items in order:
-
Verify that you have completed the API Key configuration in Step 1: Model Configuration of the initialization wizard and that the configuration succeeded. The selected region must match the region of the API key. Otherwise, the configuration fails.
-
Verify that the application is running normally. If necessary, see How do I restart the OpenClaw gateway? to restart the gateway, and then check again.
The OpenClaw CLI does not have a show subcommand for querying the token. In the OpenClaw 2026.5.19 image, running help show token returns no help topics match token. The subcommands available in the CLI of this version are reset, sandbox, secrets, security, sessions, setup, skills, status, system, tasks, terminal, tui, uninstall, update, and webhooks. Obtain the token from the Application Details page in the console. Do not try to query it by using CLI commands.
Do not share the full URL that contains the token. This link includes authentication credentials. Anyone with this link can bypass login verification and gain administrator permissions for your OpenClaw console.