Pricing
1 moveChanges seen on the public pricing pages.
-
1M CDN requests/month included, 100 GB data transfer/month included, 3 custom firewall rules, 10 IP blocking, 1M firewall rate limit requests/month included, 5K image optimization transformations/month included, 4 hours Fluid Active CPU/month included added to Hobby tier.
Product and releases
42 movesChangelog entries and releases.
-
Vercel introduced a feature allowing Chat SDK agents to stream generative UI with json-render, enabling AI-generated cards to be sent to platforms like Slack and Teams. Additionally, the AI Gateway now includes DeepSeek Search and Fetch tools, allowing users to search the web and retrieve page content using a single…
-
Vercel introduced Vercel Schedules, which is now in public beta, allowing users to run tasks on a cron schedule. Additionally, Token Pass credit will now automatically apply to model usage through AI Gateway, regardless of the authentication method used.
-
The Workflow SDK now includes idempotent starts, allowing for atomic claims of hook tokens and preventing duplicate runs. Additionally, Kimi K3.1 Pro has expanded its offerings with more Zero Data Retention providers on the Vercel AI Gateway, featuring automatic failover capabilities.
-
Vercel introduced the Vercel Security Review in public beta, which automatically reviews every push to GitHub repositories for vulnerabilities and allows the Vercel Agent to open fixes. Additionally, Vercel CI can now be used standalone, without the requirement of a linked Vercel project.
-
Data Cache is moving to Runtime Cache, which will provide faster performance under load, increased cache storage per project, and improved observability. Additionally, Vercel Daemons can now call services in their deployment through service bindings.
-
Vercel's changelog indicates that Strix now supports Vercel AI Gateway as the default inference provider, enhancing its integration with AI tools.
-
Vercel CI is now in beta, enabling users to run all continuous integration tasks, not just builds, on Vercel. Additionally, Box Mount functionality has been introduced to sync Box files with Vercel Sandbox, allowing agents to read and write files through the filesystem.
-
The new changelog entry from October 27 includes a feature where the fx harness adapter now runs the harness and model requests on the host, which improves response times and introduces new built-in capabilities.
Show 34 more
-
On 23 October, Vercel introduced Spend Management capabilities in their CLI, allowing users to manage budgets and view spending. Additionally, the Vercel MCP now features a tool for submitting feedback directly to the Vercel team.
-
Vercel announced that the free domain offer for Pro teams now includes eight more top-level domains (TLDs), such as .co, .studio, and .agency, bringing the total to sixteen.
-
Vercel's changelog update on October 21 includes the addition of Python support for routing middleware, allowing developers to write middleware using the vercel-proxy package. Additionally, Baseten's Mammoth-1 and Elephant-1 models are now available for free through Vercel AI Gateway until October 27, providing users…
-
Vercel has made WebSocket support generally available in Vercel Functions, which allows for functionalities such as multiplayer apps and chat streaming. Additionally, DeepSeek V4.1 Flash has been introduced in Token Pass, expanding the available tools for developers.
-
On 19 October, Vercel updated its Hobby and Pro Terms of Service. Additionally, they announced that Mercury 3 is now available on the AI Gateway, allowing users to call it with one API key, without needing a provider account, and including features like spend tracking and request traces.
-
Vercel AI Gateway now includes session affinity in its logs, allowing users to see grouped requests with details on cache hit rate, cost, turns, and the coding agent used. Additionally, DeepSeek V4.1 Flash 1017 is now available through the AI Gateway, featuring image input and spend tracking capabilities.
-
Vercel has added Perplexity's Decider V1.2 and Decider Flash V1.2 models to their AI Gateway, allowing access through a single API key with shared budgets and usage tracking. Additionally, Reflection AI's Beam is now available for free for a limited time on the same platform.
-
Vercel announced the introduction of Token Pass, which offers $80 in monthly credits for $30 to access featured open-weight models, starting with GLM 5.3 and Kimi K3. Additionally, Vercel Daemons is now in public beta, allowing users to run queue consumers, workflow workers, and self-hosted CI runners on the platform.
-
On October 14, Vercel announced that Amazon S3 is now available in the Vercel Marketplace with guided setup and automatic project configuration. Additionally, Vercel made Nvidia's Nemotron 3.5 Lightning available for free through the Vercel AI Gateway, requiring only one API key for access.
-
Vercel added a new feature that enables users to call Meta's Muse Wonder through the Vercel AI Gateway, simplifying access by requiring only one API key and eliminating the need for a provider account.
-
Vercel introduced a feature allowing agents to buy domains directly through the Vercel CLI. Additionally, new Pro teams will now have a default deployment retention period of 30 days, with an option for longer retention. These updates reflect ongoing enhancements to user functionality and team management capabilities.
-
Vercel introduced a feature to skip sending request bodies to Routing Middleware, which aims to reduce Fast Origin Transfer usage and improve Time to First Byte. Additionally, Vercel Sandbox creation now returns a ready sandbox, enhancing the consistency and startup time of the sandbox environment.
-
Vercel's AI Gateway now includes a feature that allows for escalation to a fallback model based on confidence conditions. Additionally, Nano Banana 2.1 has been released for generating and editing images through the AI Gateway, simplifying access with a single API key and eliminating the need for a provider account.
-
Vercel's changelog indicates that Speed Insights will no longer support the First Input Delay metric starting November 1, 2023. The Laya decision model from Convai Innovations remains available on the AI Gateway for free until October 31, 2026.
-
Vercel's changelog now includes the announcement of the Laya decision model from Convai Innovations, which is available for free through October 31, 2026. Additionally, Microsoft AI's latest audio models for speech and transcription are now accessible on the AI Gateway.
-
Vercel's CDN will no longer cache responses that include a Vary header with Cookie, which affects how certain requests are handled and logged. Additionally, Vercel Connect has been updated to accept service submissions.
-
Vercel's changelog update on 28 September includes a new feature allowing users to search for domains without needing to sign in, enhancing user accessibility. Additionally, Claude Sonnet 5.5 is now available on Vercel AI Gateway.
-
On September 25, Vercel introduced a GitHub Action for logging into their Container Registry, which enhances security by removing long-lived credentials. Additionally, Pixel Canary is now available for free on the Vercel AI Gateway.
-
Vercel Connect now supports TanStack AI, allowing for runtime authentication of TanStack AI MCP clients with managed tokens. Additionally, Vercel Blob stores are now unlimited on every plan, with store creation billed as one byte. These changes indicate a focus on enhancing integrations and storage capabilities.
-
Vercel Connect now supports Microsoft Teams, allowing agents to receive and reply to messages through the platform. Additionally, deployments now display billable duration and CPU minutes in the dashboard and other tools.
-
Vercel's changelog indicates that Enterprise teams can now set a Spend Management budget, which includes email alerts, webhooks, and the option to pause production deployments. Additionally, experimental WebMCP support has been added to the mcp-handler, allowing exposure of MCP tools to in-browser agents.
-
Vercel's Hobby plans will now automatically delete inactive deployments for accounts exceeding the 10GB limit to help users manage storage. Additionally, Secure Compute and Static IP builds will start 64% faster due to prewarmed build containers.
-
Vercel Sandbox now provides 64 GB of storage per sandbox, an increase from the previous 32 GB. Additionally, Vercel Connect allows teams on Pro and Enterprise plans to restrict who can create and manage connectors.
-
Vercel has introduced a new feature for password protection at a cost of $20 per month per protected project on the Pro plan, while Enterprise users can access it at the team level. Additionally, Vercel Authentication is now available to protect production domains for free on every plan.
-
Vercel's changelog now includes the addition of Ling 3.0 Flash Sante from Inclusion AI, available for free through October 4. Additionally, GPT 6 Astra from OpenAI is now accessible on Vercel AI Gateway for various applications including coding and research.
-
Vercel's changelog includes the new feature allowing Cursor Cloud Agents to run in Vercel Sandbox, which utilizes scale-to-zero workers and isolated microVMs. Additionally, Basic build machines with 2 vCPU and 8 GB are now available for Pro and Enterprise teams, providing a lower-cost option for build resources.
-
Vercel's Pro offer now includes free domains with the extensions .app and .dev, increasing the total eligible TLDs to eight. Additionally, the changelog notes the availability of Meta's Muse Spark 1.3 through the Vercel AI Gateway, allowing access via a single API key without the need for a provider account.
-
Vercel's changelog indicates that Qwen 3.8 Max 0902 and Claude Fable 5.1 can now be accessed through the AI Gateway using a single API key, eliminating the need for a provider account and including features like spend tracking and traces on every request.
-
'fx' is now available in the AI SDK harness layer through the @ai-sdk/harness-fx adapter, allowing users to run it via the HarnessAgent interface. Additionally, Vercel has introduced a feature that allows setting per-user budgets on the AI Gateway, enabling individual spending caps for team members alongside existing…
-
Vercel's changelog now includes a new feature allowing users to call Tencent's Hy4 Preview through the Vercel AI Gateway, which requires only one API key and includes automatic fallbacks and spend tracking.
-
Vercel introduced the ability to run Claude Managed Agents with Chat SDK, allowing users to build Slack research bots with persistent sessions and no need for Slack credentials. Additionally, Cursor is now integrated into the AI SDK harness layer, enabling its functionality through the @ai-sdk/harness-cursor.
-
Vercel's changelog indicates the addition of Meta's Muse Image and Google's Gemini 3.5 Transcribe as new features accessible via the Vercel AI Gateway. Both features allow users to call these services with a single API key, eliminating the need for separate provider accounts and providing automatic fallbacks and spend…
-
Vercel applications are now protected from recently disclosed Next.js vulnerabilities, enhancing security for users. Additionally, Speed Insights has introduced a free tier for performance overviews, while maintaining paid capabilities under Speed Insights Plus.
-
Vercel's changelog includes the announcement that Vercel Sandbox is now globally available in multiple regions including iad1, sfo1, cle1, and cdg1. Additionally, Elastic build machines will now use Turborepo cache hits to prevent tier downgrades, ensuring that heavily cached builds remain on their current machine…
In the news
18 movesCoverage in the press.
-
Vercel has introduced DBSC protection to enhance security against session cookie theft, as reported in an article by TechRepublic.
-
Vercel has launched a public beta for AI security reviews and implemented Device Bound Session Credentials to enhance security by tying login sessions to devices.
-
Vercel rolled out new protection against stolen session cookie attacks, as reported by BleepingComputer. Additionally, Vercel is beginning to implement device-bound session protection, according to Mezha.net.
-
Vercel Labs has open-sourced a tool named agent-desktop, enabling AI agents to operate using the full capabilities of a computer.
-
Vercel's CEO revealed a new AI development workflow aimed at improving software engineering processes. Additionally, the company implemented emergency firewall rules to address a critical flaw in Node.js that allowed remote code execution.
-
Vercel announced the addition of live process input for stdio MCP tools in their Sandbox, enhancing functionality. Additionally, they implemented emergency firewall rules in response to a critical flaw in the Brotli decoder that could allow remote code execution.
-
Vercel has launched a new AI model named Jev, which is reported to be the fastest-adopted model in the history of its AI Gateway. Additionally, Vercel has been involved in discussions regarding security issues related to static credentials and deployment tokens.
-
Vercel Labs announced the open-sourcing of 'fx', a Zig coding agent designed for embedding, as reported in an article on infoq.com. Additionally, Vercel implemented emergency firewall rules in response to a critical flaw in the Brotli decoder, as detailed in an article from OfficeChai.
Show 10 more
-
Vercel built a durable agent state stack, with an open-source reference app now available that operates on their AI Gateway.
-
Vercel announced that their AI model 'Jev' has become the fastest-adopted model on their platform, achieving 35% adoption among paid teams within two weeks. Additionally, Vercel reported a significant increase in AI spending as customers transition to new models.
-
Vercel introduced always-on workers to enhance backend capabilities, as reported in an article by The New Stack.
-
Vercel introduced new observability tools aimed at enhancing their web application framework, as reported in an article on DevOps.com.
-
Vercel confirmed a KVM zero-day vulnerability related to VM escape and has announced a $50,000 bounty for the researcher who reported it.
-
Vercel Labs launched scriptc, a TypeScript-to-Native compiler that operates independently of the JavaScript engine, as reported by infoq.com.
-
Vercel has implemented stricter rules for its free-tier service, addressing issues related to dormant deployments that were consuming storage.
-
Vercel has open-sourced a new library called vgpu, which is designed for AI agent shaders and is built with TypeScript. Additionally, Vercel has addressed critical bugs in their Next.js framework, as reported in a recent article.
-
Vercel launched 'Is Agentic', a free agent-readiness scoring tool that audits public websites with more than 100 checks.
-
Vercel launched a public bounty program with a reward of $1 million for individuals who can successfully escape its AI sandbox. Additionally, Vercel's Next.js framework was mentioned in relation to new features, but the primary focus was on the bounty program.
OpsControl is not affiliated with Vercel. Summaries of public sources, linked where available.