Vercel — Cloud, Developer Tools product updates and releases, tracked on megachangelog.
Hobby users can now create up to 100 Blob stores, increased from 5, providing greater flexibility to organize data by project, environment, or region. Storage, operations, and transfer limits remain in place.
Workflow SDK now integrates natively with Nitro v3, running steps in the same bundled runtime as the rest of your app instead of a separate bundle. Nitro server-side APIs like useStorage() work directly inside workflow functions, and the Workflow web UI is available at /_workflow for monitoring and debugging.
Access to Claude Fable 5 has been suspended on AI Gateway in compliance with US Government directive. Other Anthropic models remain available through the service.
Moonshot AI's Kimi K2.7 Code, a long-context coding model with multimodal (text and vision) capabilities, is now available through Vercel's AI Gateway. The model supports frontend development, DevOps, and performance optimization tasks and always runs in thinking mode.
Introduces HarnessAgent, a unified API for running multiple agent harnesses including Claude Code, Codex, and Pi. Write your agent once and swap harnesses without rewriting code, allowing you to use the best harness available at any time.
Vercel Drop enables deployment by dragging files or folders directly into the browser without Git or CLI setup. It automatically detects frameworks like Next.js for builds and publishes projects to production with a shareable live URL in seconds.
Okara, an AI CMO platform serving 120,000+ businesses, processes 4 billion tokens daily using Vercel AI Gateway for unified multi-provider model integration and Vercel Sandboxes for automated agent workflows. The infrastructure enables rapid product iteration with zero custom adapter maintenance.
The Vercel plugin is now available for Grok Build, allowing developers to access real-time Vercel knowledge including file edits and terminal commands while working. The plugin dynamically injects relevant platform APIs and recommended patterns into context.
Azure is now available as a provider for DeepSeek V4 Pro and V4 Flash models on Vercel's AI Gateway, enabling automatic failover routing and custom provider ordering without code changes. Users can bring their own Azure credentials and leverage the unified API for unified cost tracking and failover management.
Pro teams now receive partial invoices mid-cycle when on-demand usage reaches a threshold, instead of waiting until the end of the billing period. This provides clearer visibility into costs without double-billing.
Claude Fable 5 from Anthropic is now available on Vercel's AI Gateway. This advanced model excels at long-running, ambiguous, multi-step tasks with higher first-shot correctness and supports features like custom reporting, dynamic provider sorting by latency and cost, and bring-your-own-key requests without platform fees.
Set spend caps on AI Gateway API keys to control costs and prevent budget overruns. Keys can be configured with dollar limits and optional refresh periods (daily, weekly, monthly, or none) via the dashboard or CLI, applying across all providers and models.
The Vercel CLI now supports domain search functionality, allowing you to search for domain availability and pricing across all supported TLDs with filtering and sorting options. Update to CLI version 54.10.1 or later to access this feature.
Vercel's monthly AI Gateway production index reveals significant market shifts: DeepSeek V4 captured 17% of token volume in May despite launching in mid-month, while Anthropic increased its spend share to 65%. Overall token volume grew 20% and spend grew 43% month-over-month, with users increasingly routing high-volume, cost-sensitive work to cheaper models while reserving frontier models for quality-critical tasks.
Vercel Sandbox now supports persistent, attachable storage called Drives with a lifecycle independent from sandboxes. Drives can be mounted at a configurable path when starting a sandbox and remain available to attach to later sandboxes, enabling use cases like agent workspaces and retaining repositories and build outputs.
Vercel's skills.sh API is now available for querying over 600,000 open-source skills. The API uses short-lived OIDC tokens scoped to your team and project with automatic rotation, eliminating long-lived secrets, and includes a rate limit of 600 requests per minute per team and project.
Vercel updated its Terms of Service and Marketplace terms to clarify shared responsibility when AI services and third-party tools take actions on your account, added billing flexibility for non-subscription services like AI Gateway credits, and introduced Enterprise Managed Accounts with new privacy complaint handling via VeraSafe.
Nvidia's Nemotron 3 Ultra, an open Mixture-of-Experts reasoning model for agent workflows with a 1M token context window and up to 350 tokens/sec throughput, is now available through Vercel's AI Gateway. The model offers up to 30% lower cost on agentic tasks.
You can now create a Shopify store directly from Vercel and automatically configure your Shopify credentials. Build production-ready storefronts in minutes with v0, deploy without leaving your workflow, and claim the store when ready to launch.
You can now generate OpenTelemetry traces directly from the Vercel CLI using the vercel curl --trace command to trace requests to a specified endpoint, and fetch generated traces by request ID with vercel traces get. This capability is available on all plans.
Grok Imagine Video 1.5 from xAI is now available on AI Gateway, enabling video generation from input images with synchronized audio in a single pass. The model delivers improvements in audio quality, prompt following, photorealism, face accuracy, character consistency, and reference image support for greater control over visual style.
Monorepos can now configure Git settings for all deployed projects in one place instead of clicking through each project individually, making it easier to apply consistent settings like commit status and repository dispatch across multiple projects.
The Chat SDK adds first-class support for Vue and Svelte frameworks alongside React, allowing developers to build chat UIs with their framework of choice while using the same server setup. Each framework provides its own useChat helper with reactive instances for seamless template integration.
The Chat SDK now offers Slack primitives as standalone imports, allowing developers to use only needed components without loading the full runtime. This enables cleaner imports and custom implementations for apps that handle their own routing, state, or workflow execution.
Vercel Blob now supports time-bound signed URLs that allow you to generate scoped URLs for upload, download, delete, and inspect operations without exposing your entire Blob store credentials. Each URL is constrained to a single operation, pathname, and expiry (up to 7 days), and multipart upload URLs enable browsers to stream large files directly to storage.