Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- The September 21 announcement names the Beijing shared domain. New features stop being added on September 30; it does not announce shutdown of existing calls. The exact time and time zone are unpublished.
- Keep the same Workspace, region and billing plan, and copy the real API Host from the console. Example addresses are not account configuration.
- Change only the Host, retaining the corresponding /compatible-mode/v1, /api/v1 or /apps/anthropic path.
- Token Plan and Coding Plan use dedicated Hosts and Keys. Do not mechanically apply pay-as-you-go migration steps to them.
- Validate responses, streaming results and usage attribution on a small scale first. Stop expansion on anomalies. Fallback requires the old Host and capabilities to remain actually available.
The direct answer: Maintenance-mode impact and date boundaries
Alibaba Cloud announced on September 21, 2026 that dashscope.aliyuncs.com is scheduled to enter maintenance mode from September 30: no new features will be added, while existing service is unaffected. As of September 28, this is still a future planned date, not a completed switch or immediate shutdown. The announcement gives no exact time or time zone.
The current official region table maps this domain to Beijing. Singapore uses dashscope-intl.aliyuncs.com, and Hong Kong uses cn-hongkong.dashscope.aliyuncs.com. Do not expand an announcement naming the Beijing domain into a same-day shutdown claim for those two domains. Teams needing only existing capabilities can begin with an inventory; teams adopting new capabilities should prepare dedicated domains as recommended.
Define scope first: Region, Workspace, plan and inference scope
Record the current Base URL, protocol, region, Workspace, model and billing plan, and back up recoverable configuration before starting. A dedicated Host accepts same-region Keys for its target Workspace; regional endpoints, Keys and model lists must not be mixed across regions. Dedicated Token Plan and Coding Plan endpoints are outside the pay-as-you-go steps below.
A dedicated domain narrows access to the current Workspace, but switching domains does not automatically add model permissions or account quota. The region storing data at rest and the service deployment scope where inference runs are different concepts. The earlier draft's description of Virginia's -us suffix as a universal residency approach is no longer used: newer region documentation requires selecting actual deployment scope by Workspace. Base URL and region pages still describe the old US endpoint inconsistently. Check the relevant upgrade announcement and console rather than combining the pages into a new rule.
Two configuration steps: Copy the real API Host, then preserve the protocol path
First, open Workspace Management where you have access and copy the real domain from the target Workspace's API Host column. Second, replace only the request URL's Host, keeping the model, parameters and protocol path. The Beijing example format is {WorkspaceId}.cn-beijing.maas.aliyuncs.com. Replace the placeholder with your console's value; do not copy llm-xxx.
Keep /compatible-mode/v1 for OpenAI compatibility, /api/v1 for native DashScope and /apps/anthropic for Anthropic compatibility. Inventory configuration locations in SDKs, environment variables, gateways and queued tasks, then adjust in batches. Do not put keys in screenshots, logs or tickets. This article made no paid API calls and performed no account migration.
Verification and selection: Let business acceptance decide expansion
In an authorized test environment, compare new and old Hosts using the same model, parameters and low-risk input. Record time, HTTP status, request ID, returned model, streaming completeness and usage attribution. Accept only the protocols you actually need. These are suggested checks, not test results from this article, and no fixed traffic-split percentage guarantees success.
Even after short requests work, check the long requests and streaming connections used by the business. Stop expansion if the new Host cannot access the target model, authentication scope differs, response structure changes or established business thresholds are exceeded. Stop configuration changes first if the relationship between Key, plan, region and Workspace cannot be confirmed.
Failures, stopping and viable fallback conditions
For 401, first check a billing-plan/Key mismatch, then Workspace and region. If a model is missing, check the regional model list and original protocol path. Do not change Host, model, Key and network egress simultaneously in one investigation. Only when DNS, TLS, proxy 407 or connection-timeout evidence appears should you consult the systematic proxy connection troubleshooting checklist.
Keep the old configuration as a fallback option, but switch back only if the old Host, model and required capabilities still work. If you already depend on dedicated-domain features, fallback may degrade functionality or be unsuitable. No future termination date for the old service is published, so this article cannot promise indefinite availability. Changing IP does not restore permission, regional or billing eligibility.
Completion evidence and later rechecks
The completion record should include call sites, target Workspace, Host/Key ownership, protocol paths, actual acceptance results, budget and fallback conditions. Decide to switch only when this evidence is complete. A successful response does not confirm regional compliance, permissions or usage attribution.
After September 30, reread the official announcement and target-region documentation to check for plan changes and compare your request results. Actual account switching, a unified arrangement for global shared domains and compatibility of the old US endpoint remain unverified. Do not treat a documentation-review date as the platform's execution time.
Sources
Frequently Asked Questions
Will DashScope shut down on September 30?
The September 21 announcement says dashscope.aliyuncs.com enters maintenance mode with no new features and no impact on existing service. It does not announce shutdown that day; the exact time and time zone are unknown.
Will Singapore and Hong Kong shared domains also shut down that day?
The announcement names only dashscope.aliyuncs.com. Do not extend it to a same-day shutdown of dashscope-intl.aliyuncs.com or the Hong Kong domain. Check the latest announcement for your region.
Which three path types need changing during migration?
Pay-as-you-go migration changes only the Host, preserving the corresponding /compatible-mode/v1, /api/v1 and /apps/anthropic paths. Copy the actual Host from the target Workspace console.
Can Token Plan and Coding Plan use this checklist?
Not directly. They have dedicated Hosts and Keys. First confirm the billing plan, region and models you are authorized to use.
Can I always roll back when the new Host fails?
No. Confirm that the old Host and capabilities still work first. If you depend on new features, rollback may degrade functionality. Old-service availability is not guaranteed indefinitely, and an IP change is not a permission fix.