Vercel Release Notes

Workflow SDK now supports custom class serialization, letting you pass your own class instances between workflow and step functions.

Workflow SDK serializes standard JavaScript types like primitives, objects, arrays, Date, Map, Set, and more. Custom class instances were previously not supported because the serialization system didn't know how to reconstruct them. With the new @workflow/serde package, you can define how your classes are serialized and deserialized by implementing two static methods using WORKFLOW_SERIALIZE and WORKFLOW_DESERIALIZE. Here's an example of how we used custom serialization in @vercel/sandbox to greatly improve DX:

Example of how @vercel/sandbox implements workflow custom class serialization

Once implemented, instances of your class can be passed as arguments and return values between workflow and step functions, with the serialization system handling conversion automatically.

Example usage of the serialized Sandbox class within Workflow DevKit

• See the full example application utilizing @vercel/sandbox and Workflow SDK.
• Read the serialization documentation to learn more.

Qwen 3.6 Plus from Alibaba is now available on Vercel AI Gateway.

Compared to Qwen 3.5 Plus, this model adds stronger agentic coding capabilities, from frontend development to repository-level problem solving, along with improved multimodal perception and reasoning. It features a 1M context window and improved performance on tool-calling, long-horizon planning, and multilingual tasks.

To use Qwen 3.6 Plus, set model to qwen/qwen3.6-plus in the AI SDK.

AI Gateway provides a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It includes built-in custom reporting, observability, Bring Your Own Key support, and intelligent provider routing with automatic retries.

Learn more about AI Gateway, view the AI Gateway model leaderboard or try it in our model playground.

Gemma 4 26B (MoE) and 31B (Dense) from Google are now available on Vercel AI Gateway.

Built on the same architecture as Gemini 3, both open models support function-calling, agentic workflows, structured JSON output, and system instructions. Both support up to 256K context, 140+ languages, and native vision.

• 26B (MoE): Activates only 3.8B of its 26B total parameters during inference, optimized for lower latency and faster tokens-per-second.
• 31B (Dense): All parameters are active during inference, targeting higher output quality. Better suited as a foundation for fine-tuning.

To use Gemma 4, set model to google/gemma-4-31b-it or google/gemma-4-26b-a4b-it in the AI SDK.

AI Gateway provides a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It includes built-in custom reporting, observability, Bring Your Own Key support, and intelligent provider routing with automatic retries.

Learn more about AI Gateway, view the AI Gateway model leaderboard or try it in our model playground.

Chat SDK now supports Zernio, a unified social media API, with the new Zernio adapter. This is an official vendor adapter built and maintained by the Zernio team.

Teams can build bots that work across Instagram, Facebook, Telegram, WhatsApp, X/Twitter, Bluesky, and Reddit through a single integration.

Feature support varies by platform; rich cards work on Facebook, Instagram, Telegram, and WhatsApp, while editing and streaming are currently limited to Telegram.

Read the documentation to get started, browse the directory, or build your own adapter.

GLM 5V Turbo from Z.ai is now available on Vercel AI Gateway.

GLM 5V Turbo is a multimodal coding model that turns screenshots and designs into code, debugs visually, and operates GUIs autonomously. It's strong at design-to-code generation, visual code generation, and navigating real GUI environments, at a smaller parameter size than comparable models.

To use GLM 5V Turbo, set model to zai/glm-5v-turbo in the AI SDK.

AI Gateway provides a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It includes built-in custom reporting, observability, Bring Your Own Key support, and intelligent provider routing with automatic retries.

Learn more about AI Gateway, view the AI Gateway model leaderboard or try it in our model playground.

You can now transfer Marketplace resources between teams directly from the Vercel dashboard without relying on the API. This simplifies resource management during team or project changes. Both owner and member roles on the source and destination teams can initiate transfers.

The destination team must have the corresponding integration installed before receiving a resource. The feature currently supports transfer databases from Prisma, Neon, and Supabase, with additional providers and product support coming soon.

Start from your database settings in the dashboard, or learn more in the documentation.

Vercel Sandboxes can now automatically save their filesystem state when stopped and restore it when resumed. This removes the need for manual snapshots, making it easier to run long-running, durable sandboxes that continue where you left off.

How it works

A sandbox is the durable identity, now identified by a name, its filesystem state, and configuration options. A session is the compute tied to that state, invoked as needed.

Automatic persistence introduces orchestration that separates storage from compute, reducing the need for manual snapshotting, so:

• when you stop a sandbox, the session shuts down but the filesystem is automatically snapshotted.
• when you resume, a new session boots from that snapshot. This state storage is not charged, so you pay when your setting is active.

Persistence is enabled by default in the beta SDK, can be disabled between sessions with persistent: false. When disabled, the sandbox still exists after being stopped and can be resumed by its name, but each session starts with a clean filesystem.

If a sandbox is stopped and you run a command, the SDK will transparently create a new session, so you don't need to check state or manually restart

The beta SDK adds methods for managing sandboxes over their lifetime:

// Create a named sandbox
const sandbox = await Sandbox.create({ name: 'user-a-workspace' });
await sandbox.runCommand('npm', ['install']);
await sandbox.stop();
// Later, resume where you left off
const sandbox = await Sandbox.get({ name: 'user-a-workspace' });
await sandbox.runCommand('npm', ['run', 'dev']);

The beta CLI adds configuration management and session inspection:

# Spin up a sandbox for a user
sandbox create --name user-alice
# User runs commands — if the sandbox timed out, it resumes automatically
sandbox run --name user-alice -- npm test
# Check what happened across sessions
sandbox sessions list user-alice
# Tune resources without recreating
sandbox config vcpus user-alice 4
sandbox config timeout user-alice 5h

This feature is in beta and requires upgrading to the beta SDK and CLI packages.

Install the beta packages to try persistent sandboxes today:

pnpm install @vercel/sandbox@beta for the SDK, and
pnpm install -g sandbox@beta for the CLI.

Persistent sandboxes are available in beta on all plans.

Learn more in the documentation.

Vercel Sandboxes created with the latest beta package will now have a unique, customizable name within your project, replacing the previous ID-based identification. Names make sandboxes easy to find and reference:

// Create a named sandbox
const sandbox = await Sandbox.create({ name: 'user-a-workspace' });
await sandbox.runCommand('npm', ['install']);
await sandbox.stop();
// Later, resume where you left off
const sandbox = await Sandbox.get({ name: 'user-a-workspace' });
await sandbox.runCommand('npm', ['run', 'dev']);

The beta CLI adds configuration management and session inspection:

# Spin up a sandbox for a user
sandbox create --name user-alice
# User runs commands — if the sandbox timed out, it resumes automatically
sandbox run --name user-alice -- npm test
# Check what happened across sessions
sandbox sessions list user-alice

Named sandboxes are the mechanism for identifying automatic persistence, which allows your session to be more easily identified for at both time of creation resumption.

Install the beta packages to try named sandboxes today:

pnpm install @vercel/sandbox@beta for the SDK, and
pnpm install -g sandbox@beta for the CLI.

Learn more in the documentation.