Changelog

New features, improvements, behavior changes, deprecations and pricing changes on CloudBase.

CloudBase SMS package is offline, SMS verification codes with Tencent Cloud signatures are unavailable

To comply with national anti-fraud requirements and industry regulations, and to improve the security and compliance of SMS services, SMS signature requirements are tightened to match carrier standards. Starting 2026-11-13 00:00, the CloudBase SMS package is offline and SMS verification codes that use a Tencent Cloud signature are unavailable.

CloudBase environments that have used this capability must update their SMS verification code login logic in one of the following ways:

Switch to another login method, such as email verification code login or username and password login
Connect SMS with a custom signature, see Custom SMS provider configuration guide

Related documentation: SMS verification code login

Announcement: Notice on the Offline of the CloudBase SMS Package

Growth Plan image credits now support hy-image-v3.5-preview

Image generation credits from the Mini Program Growth Plan can now call hy-image-v3.5-preview. The model takes image generation requests over the Chat/Messages protocol and returns results synchronously. It supports mixed text and reference images, multi-turn editing, and up to 4096x4096 (4K) output.

Call it with ai.createImageModel("cloudbase") + generateImage, setting model to hy-image-v3.5-preview
The new model is called differently from HY-Image-3.0-Plus-4090-Tob-v1.0 and HY-Image-v3.0-I2I-ToB-v1.0.1 and is not switched automatically. You must adapt your code manually
The legacy models HY-Image-3.0-Plus-4090-Tob-v1.0 and HY-Image-v3.0-I2I-ToB-v1.0.1 remain available until 00:00 on December 1, 2026
Image URLs are valid for 12 hours, so transfer them to Cloud Storage promptly

Hy4 preview will be retired on 2026-12-07, with remaining calls switched to Hy4

The hy4-preview model will be retired on 2026-12-07 at 00:00 (UTC+8). The replacement is hy4.

If you are using model: "hy4-preview" in your code, please change the model value to hy4 before 2026-12-07. If the migration is not completed before retirement, calls will automatically switch to hy4 and be billed at hy4 rates.

Notice: Hy4 preview retirement

[Private Alpha] PostgreSQL Realtime opens to invited users

PostgreSQL Realtime is now available in Private Alpha. Browsers, Node.js, and WeChat Mini Program clients can now subscribe to database changes and states in real time using app.realtime() in the JavaScript SDK.

Realtime offers three core capabilities:

Capability	Description
Broadcast	Low-latency custom messages within a channel
Presence	Track clients in a channel and their state
Postgres Changes	Push after `INSERT`, `UPDATE` or `DELETE` on a table

Subscriptions to private channels are authorized by RLS policies on the realtime.messages table. On the management side, CLI v3.9.0 adds the tcb db pg realtime enable, get, and set commands.

Private Alpha

Available to invited users only. Apply to enable it.
APIs, parameters and console behavior may change incompatibly before general availability.
No SLA. Do not use it in production.
To apply, contact your account manager or submit a ticket.

Docs: Realtime · CLI changelog

Breaking change: tcb deploy in CLI v3.9.0 no longer builds

Starting with CLI v3.9.0 (released 2026-11-03), tcb deploy no longer runs an implicit build. When deploying an app or hosting, it only uploads existing build outputs. For scripts that rely on tcb deploy to build automatically: if there is no build output, the command fails with BUILD_OUTPUT_NOT_FOUND; if an old build output is still in the directory, it is uploaded as is, without a rebuild.

Build before you deploy. You can either run tcb app build followed by tcb deploy, or use tcb app deploy directly.

Change details

buildCommand, installCommand and framework in cloudbaserc are only used to infer the output directory. They are no longer executed.
If build output is missing, the command fails with BUILD_OUTPUT_NOT_FOUND.
Scripts parsing tcb deploy --json should note that app items in plan add a new buildAppOutputInfo field containing root and outputDir.

Docs: CLI changelog · Declarative deploy · tcb app build

[Private Alpha] Branching: create isolated copies of an environment

CloudBase has launched environment branching, allowing you to quickly spin up an isolated copy of an environment. This feature is currently in Private Alpha.

With branching, the copied environment inherits the full PostgreSQL schema, RLS policies, and existing data. You can change tables, write test data, and deploy functions on the branch without affecting the original environment. Calling CreateBranch returns the branch environment along with a temporary key that only has access to that branch. You can also manage branches using the DescribeBranchList and DeleteBranch APIs.

Private Alpha

Available to invited users only. Apply to enable it.
APIs, parameters and response fields may change incompatibly before general availability.
No SLA. Do not use it in production.
To apply, contact your account manager or submit a ticket.

Docs: Branching

Pipelines are available

CloudBase now features pipelines, providing a unified path from code to production.

You can bind a pipeline to a Git repository or an uploaded code package, and specify a target environment. Once bound, the pipeline automatically handles fetching code, installing dependencies, building, and deploying to CloudBase resources, while keeping a full history of every deployment. You can manage pipelines in the console, or call them via the CLI and API.

Docs: Pipelines overview · Console · CLI and API

AI models, October: deepseek-v4.1-flash, minimax-m3.1-flash and more added

October brings new models including glm-5.3-flashx, mimo-v2.6-pro, deepseek-v4.1-flash, and minimax-m3.1-flash. Two new models are eligible for the Mini Program Growth Plan quota, while MiniMax-M2.7 and others are scheduled for retirement.

Added

Date	Item
10-08	Models `glm-5.3-flashx`, `hy-vision-2.0-instruct`, `hunyuan-t1-vision-20250916`, `hunyuan-turbos-vision-video-20250728`, `mimo-v2.6-pro`, `mimo-v2.6-flash`, `deepseek-v4.1-flash`, `step-5-preview`
10-23	Models `minimax-m3.1-flash`, `deepseek-v4.1-flash-1017`
10-29	`hy4-preview` and `deepseek-v4.1-flash` can use the Mini Program Growth Plan quota. See the notice

Retirement announcements

Announced	Model	Notice
10-15	MiniMax-M2.7	View
10-22	Hunyuan Image 3.0 (retires 2026-12-01)	Separate entry

Full history: AI changelog.

Breaking change: new environments in the console no longer offer the document database

Starting 2026-10-30, the document database option is no longer offered when creating a new environment in the console. Environments created before 2026-10-30 are not affected; their data, collections, indexes, security rules, and SDK/HTTP API calls remain unchanged.

If you need to use document database capabilities in a new environment, choose PostgreSQL when creating the environment and enable the Document extension (pgdoc) in the console's Document editor. Once enabled, clients can continue to use the document database HTTP API and SDK with the same calls. The availability and versions of the extension depend on what the console displays.

Announcement: Document database no longer offered for new environments · Document extension

HTTP gateway routes can target a function version

HTTP gateway routes now support targeting specific versions of functions.

You can configure this directly via the "Version" dropdown in the console, or starting with CLI v3.8.10, declare it using the upstreamResourceVersion field in SCF and WEB_SCF routes. The available options are:

Value	Meaning
Version number (such as `1`, `2`)	Forward to that version. It must be published
`$LATEST`	Forward to the latest version, switching automatically on each release
`$DEFAULT`	Default. Follows the function's version traffic split, same as before

Docs: Access functions through the HTTP gateway · CLI: forward version

Hunyuan Image 3.0 models will be retired on 2026-12-01

The Hunyuan Image 3.0 text-to-image model HY-Image-3.0-Plus-4090-Tob-v1.0 and image-to-image model HY-Image-v3.0-I2I-ToB-v1.0.1 will be retired on 2026-12-01 at 00:00. Calls will fail afterward.

This retirement mostly affects users calling these models with the image quota granted by the Mini Program Growth Plan. The successor is hy-image-v3.5-preview, which uses a different calling method. Please estimate the migration work before 2026-12-01. Specific changes for the new model will be updated in the notice.

Notice: Hunyuan Image 3.0 retirement

Integrations adds Mini Program virtual payment

The Integrations center adds a "Mini Program virtual payment" integration to sell virtual goods such as in-app items, tokens, and membership subscriptions.

Once created, the integration automatically generates an HTTP function handling signing, order queries, and callbacks. The Integrations gateway verifies and decrypts callbacks before forwarding them to this function. On the Mini Program side, you can invoke wx.requestVirtualPayment to open the checkout on both Android and iOS devices.

Docs: Mini Program virtual payment

Cloud database environments can no longer enable MySQL

Starting 2026-10-20, cloud database environments no longer support enabling a MySQL database, and the entry point is closed.

MySQL can no longer be newly enabled. MySQL databases that are already enabled are not affected; you can keep using them without migrating data. Additionally, the document database in cloud database environments is not affected. If your project requires a relational database, you can create a PostgreSQL environment instead. See PG mode overview.

Announcement: MySQL can no longer be enabled

4 CloudBase experts available in WorkBuddy

The WorkBuddy expert center has added 4 CloudBase experts. Each expert covers a specific business scenario and organizes the CloudBase capabilities you need, allowing you to build apps quickly by describing your requirements.

These experts use resources in your own environment, and all generated data stays in your Tencent Cloud account. On first use, you will be prompted to connect the "Tencent Cloud CloudBase" connector to begin chatting.

Docs: WorkBuddy setup

Remote MCP supports OAuth 2.1

The CloudBase remote MCP server now supports standard OAuth 2.1 authorization. You no longer need to manually write keys in your configuration files; put the URL in the client config: China site https://tcb-api.cloud.tencent.com/mcp/v1, international site https://tcb-api.tencentcloud.com/mcp/v1. On first connection, the client will open your browser to sign in, choose an environment, and complete authorization. To switch environments, you can authorize again and pick another environment on the web page without changing your local configuration. Clients without OAuth support, and CI/CD pipelines, can keep using static credentials.

Docs: Connection modes

AI models, September: kimi-k2.8-preview added, retirement announced for 4 models

September brings new models kimi-k2.8-preview and deepseek/deepseek-flash, alongside a new model usage ranking. GLM-5 series, GLM-5v-Turbo, MiMo-V2.5-Pro, and DeepSeek-V4-Flash-0731 are scheduled for retirement.

Added

Date	Item
09-19	Models `kimi-k2.8-preview`, `deepseek/deepseek-flash`
09-28	Model usage ranking, updated from time to time with each model's share of token calls

Retirement announcements

Announced	Model	Notice
09-04	GLM-5 series	View
09-15	GLM-5v-Turbo	View
09-24	MiMo-V2.5-Pro	View
09-28	DeepSeek-V4-Flash-0731	View

Full history: AI changelog.

AI models, August: glm-5.3, hy4-preview and more added, retirement announced for 3 models

August brings new models including mimo-v2.5-pro, glm-5.3, and hy4-preview. DeepSeek-V4 models move to peak/off-peak pricing, while Kimi-K2.5, the Qwen3.5 series, and others are scheduled for retirement.

Added

Date	Models
08-03	`mimo-v2.5-pro`
08-14	`glm-5.3`
08-28	`hy4-preview`, `glm-5.3-flash`, `deepseek-v4-flash-0731`, `deepseek-v4-pro-0813`, `deepseek/deepseek-v4-flash-vision-exp`

Pricing

08-14: DeepSeek-V4 models move to peak/off-peak pricing, effective 08-17. See the separate entry.

Retirement announcements

Announced	Model	Notice
08-05	Kimi-K2.5	View
08-10	Qwen3.5 series	View
08-28	DeepSeek-V4-Flash preview	View

Full history: AI changelog.

HTTP gateway custom domains support edge acceleration

HTTP gateway custom domains now support edge acceleration. When adding a custom domain, select "CloudBase access" as the connection type to enable this feature, letting edge nodes in over 70 countries and regions serve your requests nearby.

Once enabled, you can configure cache rules for the domain, including a custom cache key. Note that this feature adds two charges: acceleration traffic and acceleration requests, billed by the region requests come from. See CloudBase pricing.

Docs: Custom domains · Cache configuration

Storage can bind multiple custom domains through the HTTP gateway

Cloud storage now supports binding multiple custom domains and serving files through the HTTP gateway.

You can bind domains directly on the "Storage - Domains" page in the console, or add an associated resource of type "Storage" in the HTTP gateway's route management. A single bucket can bind multiple custom domains, each with its own trigger path (such as / or /storage). All custom domains bound through these methods can be managed together in the HTTP gateway.

Docs: Storage custom domains · Custom domains

DeepSeek-V4 direct-supply models move to peak/off-peak pricing

Starting 2026-08-17 00:00 (UTC+8), DeepSeek-V4 direct-supply models will follow DeepSeek's official pricing policy and move to peak/off-peak pricing. Off-peak rates are only half of the peak rates.

This update affects users calling deepseek-v4-pro-202606 and deepseek-v4-flash-202605 (including active subscription orders). You do not need to make any code changes; billing is automatically calculated based on the call times:

Peak hours: 9:00–12:00 and 14:00–18:00 daily (UTC+8).
Off-peak hours: All other hours outside of the peak periods.

If you do not want to continue, you can stop using the service, release resources, or request a refund in the console before the effective date on 2026-08-17.

Unit prices: DeepSeek-V4 pricing update

CloudBase now offers PostgreSQL environments

CloudBase now officially supports PostgreSQL database environments. You can select PostgreSQL directly as the underlying database when creating a new CloudBase environment.

This mode lets you define tables, constraints, functions, and permissions using standard SQL. Once a table is created, RESTful APIs are automatically generated, ready for create, read, update, and delete actions. For security, it features a two-layer permission model: table-level GRANT controls allowed operations, and row-level RLS (Row Level Security) policies control data visibility. Additionally, extensions such as pgvector are natively supported.

Docs: PG mode overview · PostgreSQL quickstart

AI models, July: hy3 and kimi-k3 added, retirement announced for 3 models

July brings new models including hy3 and kimi-k3, along with two Hunyuan image models. MiniMax-M3 prices have dropped, while MiniMax-M2.5, DeepSeek-V3.2, and Hy3 preview are scheduled for retirement.

Added

Date	Models
07-06	`hy3`, `kimi-k2.7-code-highspeed`
07-06	Image models `HY-Image-3.0-Plus-4090-Tob-v1.0`, `HY-Image-v3.0-I2I-ToB-v1.0.1`
07-17	`kimi-k3`

Pricing

07-06: MiniMax-M3 price reduced. See the announcement.

Retirement announcements

Announced	Model	Notice
07-06	MiniMax-M2.5	View
07-06	DeepSeek-V3.2	View
07-28	Hy3 preview	View

Full history: AI changelog.

OpenAgentKernel is open source: the official CloudBase server-side Agent SDK

@cloudbase/open-agent-kernel (OAK), the official CloudBase server-side Agent SDK, is now open source under the Apache-2.0 license.

OAK includes built-in support for:

Session persistence and multi-turn context
MCP tool integration
Human approval (HITL)
Long-term user memory
Sandbox execution

To get started, make sure you have Node.js 20.19+, a CloudBase envId, and a server-side API key, then run:

npm install @cloudbase/open-agent-kernel

Docs: OpenAgentKernel · Source: GitHub

Singapore region opens on the China site, PostgreSQL environments only

China-site users can now select the Singapore region when creating an environment in the console. No separate application is needed.

The Singapore region currently offers PostgreSQL environments only; the document database and MySQL are not available. Since this region is open directly to China-site accounts, you can use your existing payment methods and console without signing up for an international-site account.

Docs: PG mode overview

CloudBase environments no longer create Lighthouse instances

Starting 2026-07-09, you can no longer create Lighthouse instances inside a CloudBase environment. To create new instances, please use the Lighthouse console.

For existing instances created inside a CloudBase environment, resource points will continue to be deducted daily to keep them available during their current validity period. If resource points run out, the instance will be isolated. You must renew it in the Lighthouse console to restore access; instances not renewed will be destroyed after the isolation period. Once the validity period ends, you must manually renew the instance in the Lighthouse console.

Announcement: Lighthouse capability removal in CloudBase

Bulk deletes in the Storage console now run as async tasks

Bulk deletes of files or folders in "Storage - Files" now run as background tasks.

When deleting in bulk, the system executes the operation in the background and opens a Task Management dialog showing real-time progress. You can close the dialog and keep working in the console while the deletion runs. Finished tasks stay in Task Management for 3 days for you to review.

Docs: Bulk delete tasks

CLI adds tcb monitor for metrics across 12 resource types

Starting with CLI v3.5.8, the new tcb monitor command allows you to query monitoring metrics directly from your terminal for 12 resource types, including functions, databases, Cloud Run, static hosting, authentication, AI models, the gateway, and environment QPS.

You can use --last 7d or --last 24h to specify the time window, or use the --json parameter to output structured results for custom scripts or CI pipelines. Running tcb monitor schema <resource> quickly lists all available metrics for that resource. To configure alarm policies, use tcb api monitor to call the cloud API.

Docs: Metrics · Alarms

CloudBase AI tools now support PostgreSQL, plugin marketplace, and environment management

CloudBase AI development tools introduce several updates, including PostgreSQL database support, plugin ecosystem integration, authentication enhancements, and self-service environment management.

Plugin ecosystem

CloudBase plugins are live, allowing installation into mainstream AI coding assistants via the marketplace:

Codex / Claude Code: Add the CloudBase marketplace and install. See the installation guide
WorkBuddy: Built-in CloudBase connector, enable directly in the chat interface

Authentication

Added API Key authentication, supporting project-level credentials for cloud IDEs, CI/CD, and automation.

PostgreSQL support

Added PostgreSQL MCP tool set and schema-first workflow, supporting full PostgreSQL database operations.

Self-service environment management

Query plans, create environments, change plans (upgrade/downgrade), and renew directly via the tools.

Integration Center is live with WeChat Pay and Official Account support

CloudBase has launched the Integration Center to package third-party services like WeChat Pay and WeChat Official Account into integrations. You no longer need to manage credentials, handle signature verification, decrypt data, or maintain callback domains. Fill in the configuration once in the console to generate cloud functions and callback pipelines. Feature introduction

Supported integrations

WeChat Pay: Mini Program Payment / JSAPI Web Payment / Native QR Code Payment
WeChat Official Account: Official Account Open Services (web authorization, user messages, template messages, custom menus, etc.)

Features

Managed credentials: API keys, certificates, and private keys are managed by the platform and never enter business code
Callback verification: Automatically receives async notifications from third parties, verifies and decrypts them, and forwards plaintext to business functions
Function deployment: Automatically generates HTTP cloud function templates when creating integrations; you only need to fill in business logic

Related docs: Integration Center Documentation · Usage Guide · FAQ

CloudBase environments are now eligible as ICP filing cloud resources

CloudBase environments on Personal plans or above with more than 6 months of remaining validity and Fixed IP enabled can now be directly retrieved in the filing console as eligible cloud resources. You do not need to purchase additional cloud services for filing.

How to file

Go to Filing Management
If filing conditions are not met, follow the on-page instructions.
If conditions are met, click "Go to Filing" at the top of the page to enter the filing console.

On the filing page, eligible CloudBase environments can be retrieved as filing cloud resources.

Enter the filing application process and follow the ICP Filing Documentation for further steps.

CLI, MCP Server, and AI Skills now support Device Code Flow

CloudBase CLI, MCP Server, and AI Skills now support Device Code Flow login. This feature is designed for headless environments or remote servers (such as cloud IDEs, SSH terminals, and OpenClaw), allowing identity authorization without a local browser.

Sign-in for remote development

Traditional login methods rely on a local browser redirect to complete OAuth authorization, which cannot be used directly in remote servers, container environments, or cloud IDEs. Device Code Flow decouples the authorization device from the login device through a device code. You only need to visit the authorization page in any browser and enter the device code to authorize the remote environment.

Use with AI agents

When using CloudBase Skills inside AI development tools (such as Cursor, WindSurf, CodeBuddy, etc.), the Agent automatically triggers the Device Code Flow. Once authorized, it can invoke the full set of CloudBase capabilities, enabling automation from development to deployment.

Example: an AI agent in OpenClaw

Using an OpenClaw cloud development environment as an example:

Step	Role	Action
1	User	Sends instruction to Agent: `Use CloudBase Skills, build a TodoList web app and deploy it`
2	Agent	Auto-discovers and installs CloudBase Skills, triggers Device Code Flow
3	Agent	Returns authorization URL and device code (e.g., `FSN9-9B5D`), waits for user to authorize
4	User	Opens the URL in any browser and enters the device code to complete login
5	Agent	Detects authorization success, continues automatically: scaffolds project → writes code → calls CloudBase API → deploys

Redesigned Cloud Storage console supports COS content moderation and data processing

CloudBase Cloud Storage console has been redesigned to display COS bucket names and region information directly, and now supports COS advanced features such as content moderation and data processing.

What's new

Bucket details: Directly displays COS bucket name and region information
Open in COS: Quick jump to COS console to use advanced features
Content moderation: Image, video, and text security auditing to automatically identify non-compliant content
Data processing: Image processing, document preview, video frame capture, transcoding, and more

Related docs: Cloud Storage Documentation

HTTP Cloud Functions support multi-concurrency and WebSocket real-time communication

HTTP Cloud Functions now support single-instance multi-concurrency mode and WebSocket/SSE real-time communication. Development Documentation

Multi-concurrency reduces cold starts

A single function instance can handle multiple requests simultaneously, removing the need to start a new instance for every request. Compared to the "single instance single request" mode of regular cloud functions, it significantly reduces cold start frequency and resource consumption.

Bidirectional WebSocket communication

Native support for WebSocket protocol provides full-duplex communication. This is ideal for scenarios requiring bidirectional communication, such as real-time chat, collaborative editing, and game servers. WebSocket Usage Guide

MySQL databases can now migrate to your own Tencent Cloud account

CloudBase MySQL now supports migrating from the CloudBase side to your Tencent Cloud account. After migration, it supports VPC private network connections, allowing cloud functions to connect directly to the database. Migration Guide

Before this, the database was deployed in the official CloudBase account and only supported public network connections. After migration, the database is under your account, and resources like cloud functions and CloudRun can connect via the private network.

Console overview page is upgraded to Task Center

The CloudBase platform overview page has been upgraded to the Task Center. This unifies the management of environment risks, security reminders, and operational anomalies scattered in various places, allowing you to identify and handle all to-do items in one place.

Manage environment to-do items in one place

CloudBase launches resource point billing mode

CloudBase has launched the resource point billing mode, replacing the traditional "package quota + additional resource" fixed quota method to enable flexible deduction based on actual usage. This mode covers the full range of current packages, including Personal, Standard, Enterprise, and Enterprise Advanced versions. Resource Point Billing Documentation

Pay for what you use

All resource usage is converted to resource points based on unified standards and deducted accordingly. You only pay for what you use.

Important

WeDa/CloudBase discontinued old packages need to be converted to current packages (Personal, Standard, Enterprise, Enterprise Advanced) first before converting to resource point billing environments

CloudBase now supports connecting to your own MongoDB databases

CloudBase has added the Mongo Connector feature, supporting the integration of your existing MongoDB database into the platform without data migration. Reuse existing resources with a configuration time of less than 5 minutes.

After connecting, you can

No migration: Directly manage data in your existing databases without data migration
Data models: Quickly generate data models based on existing data tables for automated data management
SDK access: Unified access using CloudBase Web SDK / Node SDK / Mini Program SDK
Data control: Data remains in your database, giving you full control over data security

Default domain renewal retired in favor of access warning page

The original manual default domain validity renewal mechanism has been officially retired, removing the need for frequent manual renewals. An access warning intermediate page has been introduced, replacing the hard renewal requirement with soft security reminders to ensure development and testing environments remain secure and compliant.

When visitors access Web applications through CloudBase default domains (such as *.tcloudbaseapp.com, *.run.tcloudbase.com, *.app.tcloudbase.com), the system automatically redirects to the access warning intermediate page. This informs visitors that the current environment is only for development and testing, and the content has not undergone compliance review.

From mandatory renewal to a warning page

In the old mechanism, default domains required regular manual renewals. If forgotten, domains became inaccessible, interrupting development and testing. The new mechanism removes the renewal requirement. As long as traffic anomaly risk control thresholds are not triggered, default domains can continue to be used in development and testing scenarios.

Related docs: Access Warning Intermediate Page

Added framework quick access guides and AI tool MCP connections

CloudBase has added an Access Guide module, providing quick access solutions for mainstream frameworks such as React, Vue, and WeChat Mini Programs, along with MCP (Model Context Protocol) connection guides for AI tools like Cursor and WindSurf.

Framework quick start guides

You no longer need to consult multiple documents and manually configure SDKs. Simply select your technology stack in the guide to get complete project templates and configuration solutions.

MCP connection for AI tools

To improve AI-assisted development efficiency, CloudBase provides MCP connection guides for mainstream AI tools, allowing AI assistants to directly call CloudBase capabilities.

CloudBase AI development tools have expanded to support 20+ AI tools (including WorkBuddy, Codex, Claude Code, and more), plus one-click plugin installation — no manual MCP configuration required.

Authentication features zero-configuration email service and SMS rate limiting

CloudBase authentication has added a zero-configuration email service and introduced SMS rate limiting to prevent SMS resource abuse.

Built-in email service

You no longer need to apply for third-party email services or configure complex SMTP parameters. Simply enable "Enable Email Proxy" in the console to complete email verification code login configuration.

SMS rate limiting

To prevent cost spikes caused by malicious SMS verification code requests, the system has added a daily SMS limit configuration per user (default 30 messages/day).

Configuration steps:

Go to CloudBase Platform / Authentication / Login Methods
Find the "SMS Verification Code" card and click "Edit"
Set the daily sending limit (recommended 10-100 messages)
Save the configuration

Important

The system has enabled default rate limiting (30 messages/day) for all environments. If your business scenario requires higher quotas (such as order notifications for e-commerce platforms), please manually adjust in the console to avoid affecting normal business.

Cloud Storage and Static Hosting support viewing COS buckets

CloudBase Cloud Storage and Static Hosting have added a feature to view COS buckets, supporting direct redirection to the underlying COS management console to use advanced storage and management capabilities.

Open the COS console directly

Click "View COS Bucket" on the Cloud Storage or Static Website Hosting page to jump directly to the corresponding COS bucket management page. This allows you to use advanced capabilities like content moderation, data processing, and access logs without manually constructing access routes.

Full notices for deprecations and pricing changes: Announcements
Version history of each tool: CLI · JS SDK · Node.js SDK · Manager SDK · AI

Change details​

Added​

Retirement announcements​

Added​

Retirement announcements​

Added​

Pricing​

Retirement announcements​

Added​

Pricing​

Retirement announcements​

Plugin ecosystem​

Authentication​

PostgreSQL support​

Self-service environment management​

Supported integrations​

Features​

How to file​

Sign-in for remote development​

Use with AI agents​

Example: an AI agent in OpenClaw​

What's new​

Multi-concurrency reduces cold starts​

Bidirectional WebSocket communication​

Pay for what you use​

After connecting, you can​

From mandatory renewal to a warning page​

Framework quick start guides​

MCP connection for AI tools​

Built-in email service​

SMS rate limiting​

Open the COS console directly​

Change details

Added

Retirement announcements

Added

Retirement announcements

Added

Pricing

Retirement announcements

Added

Pricing

Retirement announcements

Plugin ecosystem

Authentication

PostgreSQL support

Self-service environment management

Supported integrations

Features

How to file

Sign-in for remote development

Use with AI agents

Example: an AI agent in OpenClaw

What's new

Multi-concurrency reduces cold starts

Bidirectional WebSocket communication

Pay for what you use

After connecting, you can

From mandatory renewal to a warning page

Framework quick start guides

MCP connection for AI tools

Built-in email service

SMS rate limiting

Open the COS console directly