This topic describes the release notes for Volcano Engine Function as a Service (veFaaS). New features will be rolled out across regions. We are looking forward to your hands-on experience.

August 2026

Feature	Description	Region	Related document
Cloud sandbox applications support custom DNS	You can configure custom DNS servers and host resolution rules for sandbox applications to meet requirements such as custom DNS resolution within sandbox instances and access to internal network services.	All	Advanced configurations for sandbox applications

June 2026

Feature	Description	Region	Related document
Suspending and resuming sandbox instances	[Beta - apply for trial] You can suspend (PauseSandbox) a running sandbox instance to save its state, release underlying resources, and stop billing. You can later resume (ResumeSandbox) the instance from the snapshot to restore its state and continue running, enabling on-demand suspension and reducing idle costs.	All	Pause and resume sandbox instances
Sandbox applications support the E2B protocol	[Beta - apply for trial] The cloud sandbox now supports the E2B protocol at the API layer and allows injecting the Envd runtime via Sidecar into any sandbox image. You can use the official E2B CLI/SDK to create, connect, execute commands, read/write files, and run code in sandbox instances without modifying your business code.	All	Sandbox applications integrate with the E2B protocol

March 2026

Feature	Description	Region	Related document
Cloud sandboxes supporting route authentication configuration	Cloud sandboxes support configuring route authentication. Users can use the Key Auth plugin through APIG to implement API Key-based authentication and authorization. This ensures the route security of sandbox applications.	All	Sandbox application
Cloud sandboxes supporting warm pools	Through a unified product solution, veFaaS offers sandbox instance warm pool capabilities, enabling centralized governance and observability of the warm pool and significantly reducing instance startup duration.	All	Sandbox instance warm pools
Cloud sandboxes supporting application-level QPM	[Beta - apply for trial] Cloud sandboxes support application-level queries per minute (QPM) of the instance creation API. This ensures that high-priority sandbox applications obtain stable and sufficient QPM resources and prevents other applications from occupying resources.	All	Sandbox application

December 2025

Feature	Description	Region	Related document
Application deployment through code repositories and code packages	Applications can be deployed through remote code repositories or local code packages.	All	Managing applications
Sandbox image filtering by type	Public sandbox images can be filtered by type, which improves filtering efficiency.	All	For sandbox images
Custom runtime	[Beta - apply for trial] You can create runtimes based on custom images to meet specific business requirements.	All	Creating custom runtimes

November 2025

Feature	Description	Region	Related document
CLB support for microservices apps	Microservices apps allow you to specify a CLB instance as the traffic entry point.	All	Adding CLB access configurations
Wildcard domains supported for sandbox applications	[Beta - apply for trial] In the sandbox environment, a service contains multiple sandboxes and shares a public access address. Through a wildcard domain, each sandbox can be assigned a unique subdomain to implement independent access.	All	Wildcard domain configuration for sandbox applications
Automatic application builds and deployments based on code repositories	Applications can be associated with user-defined code repositories. The application deployment pipeline is automatically triggered by code repository events.	China (Beijing) China (Guangzhou)	Creating applications that support automatic deployment based on code repositories

September 2025

Feature	Description	Region	Related document
Management optimizations for sandboxed applications	Sandbox applications support manual release, rollback, and version management.	All	Sandbox application
Health check	Microservices functions support health checking.	All	Function configuration
Debugging	In native runtime scenarios for web apps, container debugging supports the WebIDE debugging method.	All	Debugging
New application templates	The veFaaS Application Center supports one-click deployment of All-in-One Sandbox and Gem Browser applications.	All	Deploying All-in-One Sandbox applications with a few clicks Deploying Gem Browser applications with a few clicks

July 2025

Feature	Description	Region	Related document
Agents are now available in the App plaza.	New applications, including Coze Studio and Coze Loop, are now available in the App plaza of the veFaaS console. You can achieve the private deployment of the open-source Coze platform on the cloud with a few clicks by using veFaaS.	China (Beijing)	Deploying the Coze Studio application with a few clicks Deploying the Coze Loop application with a few clicks
Price reductions are applied to some billing items.	Price reductions are applied to some billing items in the Asia Pacific (Johor) region, such as the number of function invocations and CPU instance resource usage.	Asia Pacific (Johor)	[Billing updates] Notification on the changes in veFaaS partial unit prices
The configuration of the asynchronous task feature is optimized.	When the asynchronous task feature is in non-webserver mode, the exclusive mode is used by default, and a single instance cannot process multiple concurrent requests. This improves the success rate of asynchronous task creation.	All	Configuring asynchronous tasks Function configuration
Cloud sandbox is added.	The serverless cloud sandbox capability is added to provide a highly elastic, cost-effective, secure, and isolated task execution environment for AI inference and training.	All	veFaaS cloud sandbox management
Agents are now available in the App plaza.	New applications, including Browser Session, Long-term Memory Agent, and Mobile Use Agent, are now available in the App plaza of the veFaaS console. You can experience the applications at no cost and create an application with a few clicks. The Agent application scenarios are expanded.	China (Beijing)	Deploying Browser Session applications with a few clicks Introduction to Long-term Memory Agent Deploying Long-term Memory Agent applications with a few clicks Introduction and quick deployment guide of Mobile Use Agent

April 2025

Feature	Description	Region	Related document
App plaza allows you to experience Agent apps and create the apps with a few clicks.	The App plaza is now available, providing agent applications including Browser Use Agent, Code Sandbox Agent, and Computer Use Agent for you to try free of charge, with some applications supporting one-click creation. It offers a general Agent solution that simulates completing open-ended tasks across various OS environments, helping enterprises and individuals build their own general-purpose Agents and enriching Agent application scenarios.	China (Beijing)	App plaza AI Cloud Native
The Python and native Python runtimes can be integrated with the Application Performance Management Plus (APMPlus) monitoring service.	The monitoring data of functions is delivered to APMPlus, which provides integrated monitoring for metrics, traces, and logs, creates monitoring dashboards, and monitors high-code applications on ModelArk as well as intelligent agents such as LangChain.	China (Beijing) China (Shanghai)	None

March 2025

Feature	Description	Region	Related document
You can specify a vCPU-to-memory ratio for a CPU instance.	You can select an appropriate combination of vCPUs and memory for a CPU instance based on different business requirements. In addition, for the new combination, the original memory usage is split into vCPU usage and memory usage, with unit prices unchanged. This enriches CPU instance specifications to meet the requirements of diverse business scenarios.	All	Function configuration Product announcement Pay-as-you-go

February 2025

Feature	Description	Region	Related document
Asynchronous tasks support callback notifications.	Asynchronous tasks support result callback notifications, which enable downstream services to be notified of the execution results of asynchronous functions. The asynchronous task feature is improved to apply to more application scenarios.	All	Asynchronous task result callback
Asynchronous tasks can be retried.	You can specify the number of retries for asynchronous tasks. If a task fails to be executed, it can be retried without affecting the overall process. The asynchronous task feature is improved to apply to more application scenarios.	All	Configuring asynchronous tasks
The limits on asynchronous task records are changed.	The retention period for asynchronous task execution status is extended from the original 24 hours to approximately 30 days, or up to 1,000 status records. Storing more execution records provides you with comprehensive monitoring metrics and log observability.	All	Configuring asynchronous tasks Monitoring tasks
A Debug instance can be created before a function is released.	Before you release a function, you can create a Debug instance to perform remote debugging. This optimizes the debugging procedure when a function is created for the first time, in which debugging and testing are performed on the function before it is released.	All	Debugging
imageconfig can be parsed to read execution commands and environment variables.	You can resolve the imageconfig file during the creation and configuration of image functions to read and directly configure execution commands and environment variables. This optimizes the image function creation process and user experience.	All	Creating web apps Create tasks Managing functions
Environment variables can be edited in JSON format and copied in batches.	The environment variables of function instances can be edited in JSON format and copied in batches. This optimizes the operations related to function creation and management, and provides multiple methods for configuring environment variables.	All	Function configuration

December 2024

Feature	Description	Region	Related document
Web apps support debugging through Debug instances.	You can perform remote debugging on web apps that use the native runtime by using Debug instances. A Debug instance is a cloud-based instance with a custom lifecycle for temporary testing, eliminating the cumbersome cycle of repeated modification, release, and verification.	All	Debugging

October 2024

Feature	Description	Region	Related document
You can enable a single instance to process multiple concurrent requests for asynchronous tasks.	The feature of enabling a single instance to process multiple concurrent requests is added for synchronous tasks. After you enable this feature, a single instance can process multiple concurrent requests.	All	Function configuration
The MQ consumer function template is added.	The Message Queue (MQ) consumer function template is added under the microservices app category, which allows you to deploy and start a data processing service with a few clicks to synchronize the data from MQ to Cloud Search (ES).	All	Synchronizing Kafka data to ES by using veFaaS
Asynchronous tasks support custom startup commands.	You can specify a custom startup command when you submit an asynchronous task each time to meet the requirements of diverse task execution scenarios.	All	Submitting tasks
Projects and tags can be configured for functions.	You can group and manage function resources by project and tag. In this way, you can perform operations on resources in groups such as Identity and Access Management (IAM) authorization and bill splitting.	All	Project management Label management Best practices on IAM authorization
Scheduled scaling policies can be configured for functions.	When a function exhibits obvious periodic patterns or predictable traffic spikes, you can configure scheduled policies based on business characteristics to implement scheduled scaling of instances. This improves instance utilization and addresses resource waste.	All	Function resource statistics Configure the scheduled scaling policy for instance numbers
The application center provides the Mirror Maker data synchronization application.	[Beta - apply for trial] The application center provides the volcano-mirror-maker application template, which hosts the Mirror Maker 2.0 microservices app and supports data synchronization between Kafka clusters.	All	Deploying a Mirror Maker data sync application with a few clicks

September 2024

Feature	Description	Region	Related document
Microservices app hosting	veFaaS has launched the microservices app hosting feature for hosting microservices apps in the CPU always allocated mode. You can directly host container applications without the need to integrate microservices apps with triggers.	All	Microservices app

August 2024

Feature	Description	Region	Related document
The Node.js runtime supports online dependency installation.	The Node.js runtime supports online installation of dependencies, which simplifies the procedure and prevents errors that may occur during local installation, packaging, and upload of dependencies.	All	Node.js dependency installation
The native Python runtime is added.	Native Python 3.8, 3.9, and 3.12 runtimes are added, which enable you to develop web apps by using native Python HTTP frameworks without the need to install third-party interpreters.	All	Native Python runtime
A code package can be uploaded when a function is being created.	When you create a function, you can upload a ZIP package from your on-premises device or through a TOS bucket, which simplifies the procedure.	All	Creating functions

July 2024

Feature	Description	Region	Related document
The asynchronous task feature supports non-webserver mode.	The non-webserver mode is added for the asynchronous task feature, which allows you to execute scripts by using tasks without the need to implement an HTTP server. The process will automatically terminate if the execution is complete or times out.	All	Create serverless task hosting
You can specify custom startup commands and ports for serverless application hosting.	Custom startup commands and ports can be configured for serverless application hosting, which eliminates the need to write the run.sh startup script and helps you smoothly migrate native web framework applications.	All	Creating serverless application hosting
The Python runtime supports online dependency installation.	The Python runtime supports online installation of dependencies, which simplifies the procedure and prevents errors that may occur during local installation, packaging, and upload of dependencies.	All	Python dependency installation
The native runtime is officially available.	The native runtime is a custom runtime environment provided by veFaaS, which can be deployed through a code package and a container image. You can create a dedicated function execution environment by using the native runtime.	All	Native function development guide
The asynchronous task feature is officially available.	The asynchronous execution mode can be used to respond to invocation events and track and monitor task status. This feature is suitable for long-running or compute-intensive tasks, which can address issues such as continuous occupation of caller resources during synchronous invocation and meet the requirements for high stability of traces.	All	Asynchronous tasks
The serverless application center is officially available.	This feature aggregates function instances, triggers, and other dependent cloud resources, which helps you deploy serverless applications with a few clicks and achieves the full-lifecycle management of applications.	All	Application center
The Express.js web service application template is added.	This feature helps you deploy Express.js web services with a few clicks.	All	Deploying an Express.js application with a few clicks

June 2024

Feature	Description	Region	Related document
Prepaid resource packages are added.	veFaaS has launched prepaid resource packages, which can be used to offset different billing items. This billing method is more cost-efficient than pay-as-you-go.	All	Resource package
GPU functions support reserved instances and billing.	Static reserved instances can be configured for GPU functions. As a result, reserved instances can be configured for both CPU functions and GPU functions.	All	Reserved instance Billing overview
A function can be configured to access the public network.	After VPC access is enabled for a function, users can specify how the function accesses the public network.	All	Function configuration
The TLS trigger is added.	Torch Log Service (TLS) is added as one of the event sources for veFaaS, which enables you to write functions to perform extract, transform, and load (ETL) processing on massive logs.	All	Creating TLS triggers

May 2024

Feature	Description	Region	Related document
The serverless application center is added.	[Beta] This feature aggregates function instances, triggers, and other dependent cloud resources, which helps you deploy serverless applications with a few clicks and achieves the full-lifecycle management of applications.	All	Application center
Static reserved instance support for CPU functions	Configuring static reserved instances effectively prevents request latency caused by cold starts of elastic instances during peaks of function invocation. warning You will be charged for reserved instances. Regardless of whether a request is executed, the memory usage fees of reserved instances will be continuously generated until the instances are released.	All	Reserved instance Billing overview
Outbound Internet traffic is billed on a pay-as-you-go basis.	veFaaS starts to charge you for outbound Internet traffic routed through the default network interface card (NIC). The traffic is billed on a pay-as-you-go basis. Make sure that your account balance is sufficient to prevent your business from being disrupted.	All	Billing overview
Capacity FileNAS can be mounted.	You can configure Capacity FileNAS for functions, which provides more options to meet the requirements of different business scenarios.	All	Mounting FileNAS

March 2024

Feature	Description	Region	Related document
TOS can be mounted.	You can configure Torch Object Storage (TOS) for a function. You can specify a mapping relationship between a bucket subdirectory and the local directory of the function to grant the read and write permissions on the bucket subdirectory to the function. This solution overcomes the limitations of local disk space, which achieves persistent data storage and sharing among multiple functions.	All	Mount TOS Using veFaaS to read and write the TOS mount directory
The logic for retrieving asynchronous task logs is optimized.	You can identify the complete logs for each task by using the instance name and the start and end times of the task. In this case, the request ID is not required.	All	Monitoring tasks

January 2024

Feature	Description	Region	Related document
Extreme FileNAS can be mounted.	You can configure Extreme FileNAS for functions. You can specify a mapping relationship between a NAS remote directory and the local directory of the function to grant the read and write permissions on the NAS file system to the function. This solution overcomes the limitations of local disk space, which achieves persistent data storage and sharing among multiple functions.	All	Mount FileNAS Using Function as a Service to read and write NAS file system Quickly building a visual file manager for NAS by using veFaaS
New GPU metrics are added.	New GPU-related monitoring metrics are added, including function GPU utilization, function GPU video memory usage, and function GPU video memory utilization.	China (Shanghai)	Dashboard overview
Release management is added.	The function release process can be monitored in real time. For example, traffic proportions and instance status can be monitored. The release notes can be managed. For example, you can view release details and roll back to historical versions. The version records can be managed. For example, you can view the configuration and code of each version and overwrite the current version with the latest version.	All	Release observation Release records Version history
The creation process is optimized.	Based on different function programming models, functions can be created though built-in runtimes and serverless application hosting. The function creation process and interaction design are optimized, which enhances availability.	All	Creating functions

November 2023

Feature	Description	Region	Related document
The serverless GPU service is added.	[Beta - apply for trial] veFaaS launches the Serverless GPU service that continues the principle of serverless computing, which supports the efficient and cost-effective deployment of AI applications.	China (Shanghai)	Serverless GPU
Deployment across multiple availability zones is supported.	Function instances can be deployed across multiple availability zones to ensure business high availability.	All	Creating functions
The BMQ trigger is added.	The cloud-native message engine BMQ is added as one of the event sources for veFaaS, which increases the types of MQ triggers.	All	Creating BMQ triggers

September 2023

Feature	Description	Region	Related document
Asynchronous tasks are supported.	[Beta] The asynchronous execution mode can be used to respond to invocation events and track and monitor task status. This feature is suitable for long-running or compute-intensive tasks, which can address issues such as continuous occupation of caller resources during synchronous invocation and meet the requirements for high stability of traces.	All	Asynchronous tasks

April 2023

Feature	Description	Region	Related document
The compatibility of the APIG trigger is optimized.	The upstream capabilities of API Gateway (APIG) are adapted and the compatibility of the APIG trigger is improved. This decouples routing from functions and flexibly supports scenarios such as multi-version and multi-environment management of functions.	All	Creating APIG triggers

December 2022

Feature	Description	Region	Related document
veFaaS	veFaaS is released for the first time, which provides a serverless computing service to enable enterprises to quickly create various applications at low costs.	All	What is veFaaS?