timace
Releases

Vercel releases

Every Vercel update, newest first: AI Gateway models, Sandbox, Functions, the CLI, pricing and platform releases.

November 2026

  1. Launch

    Gemini Decision 1 now available on Vercel AI Gateway

    Vercel announced that Gemini Decision 1 from Google is now available on AI Gateway. The multimodal decision model accepts text, images, audio, video, and PDF inputs and returns Boolean probabilities, choices, and scores without generating output tokens.

  2. Launch

    MiniMax H3 Image and H3 Image Highspeed now available on Vercel AI Gateway

    Vercel announced that the MiniMax H3 Image and MiniMax H3 Image Highspeed models are available through AI Gateway. Both models support text-to-image generation and image editing, with custom dimensions and aspect ratios including 2K and 4K outputs; the Highspeed variant offers faster generation with the same options.

  3. Update

    Vercel Speed Insights and Web Analytics now attribute data to custom environments

    Vercel updated Speed Insights and Web Analytics so that data from deployments to custom environments is attributed to the custom environment instead of the Preview environment. Data can be viewed in the dashboard Environment dropdown, via the vc metrics CLI command, in the query builder, and through the Aggregation API (Web Analytics only).

  4. Update

    Vercel Sandbox snapshot creation is now up to 10x faster

    Vercel Sandbox snapshot creation is now up to 10x faster, with a median time to snapshot under 0.4 seconds, down from 4.4 seconds. Snapshot restore speed improved 50% at p50, from 500ms to 250ms, via a new indexing approach in the Vercel Hive Snapshot (VHS) format.

  5. Model release

    Claude Fable 5.2 now available on Vercel AI Gateway

    Vercel announced that Anthropic's Claude Fable 5.2 is available on AI Gateway. The model supports effort levels low, medium, and high, with high as the API default, and has cybersecurity and biology safety classifiers enabled.

  6. Model release

    Muse Ember now available on Vercel AI Gateway

    Vercel announced that Muse Ember, a lightweight reasoning model from Meta, is now available on AI Gateway under the model name meta/muse-ember. The model accepts text, image, and video input with a 1M-token context window, supports tool calling, and has adjustable reasoning effort from minimal to max.

  7. Update

    Workflow SDK v5.2.0 writes events and streams over WebSockets on Vercel by default

    The Workflow SDK now sends event writes and stream writes over a WebSocket by default, replacing one HTTP request per write (or per write group for streams). No code change is required beyond upgrading to workflow@5.2.0 or later. In a benchmark replaying an AI agent's token stream, v5.2.0 delivered the full response 4.2× faster than v4.8.12, with no chunk delayed over a second.

  8. Model release

    GLM 5.3 Flash is now available in Vercel Token Pass

    Vercel announced that Z.ai's GLM 5.3 Flash is now available in Token Pass, joining Kimi K3, GLM 5.3, and DeepSeek V4.1 Flash. GLM 5.3 Flash is described by Vercel as Z.ai's native multimodal coding model with a 1M-token context window.

  9. Update

    Vercel build disk space increased to 64 GB on Basic and Standard machines

    Vercel increased build disk space from 32 GB to 64 GB on Basic and Standard build machines, matching the capacity already available on Enhanced and Turbo machines. The change applies automatically across all plans and build machine types at no additional charge, with no configuration changes required.

  10. Launch

    DeepSeek V4.1 Flash 1017 Ultrafast now available on Vercel AI Gateway in public preview

    Vercel announced that DeepSeek V4.1 Flash 1017 Ultrafast, the October 17 update of DeepSeek V4.1 Flash served through Fireworks' accelerated serverless endpoint, is available on AI Gateway in public preview. The model has a 1M-token context window and is accessible with no waitroom or session time limit, using the model name deepseek/deepseek-v4.1-flash-1017-ultrafast.

  11. Update

    Vercel Token Pass credit now applies automatically across AI Gateway

    Vercel updated Token Pass so that its credit applies automatically to included model usage in AI Gateway, for requests authenticated with an API key, OIDC token, or login, without requiring a dedicated Token Pass key. When Token Pass credit runs out for the billing period, standard keys and OIDC tokens fall back to the team's AI Gateway credits, while Token Pass keys stop until the next billing period.

  12. Update

    Vercel AI Gateway adds DeepSeek Search and Fetch tools

    Vercel AI Gateway now offers DeepSeek Search and DeepSeek Fetch tools. Search returns relevant pages with titles, URLs, and summaries; Fetch retrieves page content as Markdown for a model to read. Both are priced at $0.75 per 1,000 billable requests, with model tokens billed separately.

  13. Update

    Vercel Chat SDK adds streaming generative UI via json-render

    Vercel's changelog announced that Chat SDK agents can now stream generative UI using json-render. An agent writes a JSON spec built from a typed component catalog, which is converted into a Chat SDK card and posted in each platform's native format, such as Block Kit on Slack and Adaptive Cards on Teams. The stream updates the message at most once every 500 ms by default, and a validation function checks element props and platform limits.

  14. Update

    Vercel CLI adds option to set new projects to expire

    Vercel CLI now supports a --delete-project-after flag on vercel deploy, which sets a newly created project to be deleted automatically after a chosen duration. Supported durations are 1h, 1d, 1w, 2w, and 30d, and the CLI prints the expiration date and time; running vercel project keep removes an expiration, and the feature requires Vercel CLI v65.12.0 or later.

  15. Launch

    Vercel Tracing is now generally available

    Vercel Tracing reached general availability on all plans. The new Traces page in the dashboard and the vercel traces CLI commands allow searching and inspecting traces with KQL, and new Next.js projects get tracing enabled by default after their first successful build.

  16. Model release

    Ampersand from Humans& now available on Vercel AI Gateway

    Ampersand, a language model from Humans&, is now available on Vercel AI Gateway with model name humansand/ampersand-v0.1, free through November 13, 2026. The model is designed for group conversations and decides on each turn whether to speak or stay quiet, using caller-defined send_message and pass function tools.

  17. Launch

    Gemini Omni 1.2 Flash now available on Vercel AI Gateway

    Vercel announced that Google's Gemini Omni 1.2 Flash is available through Vercel AI Gateway. The model generates 3–15 second videos with native audio from text prompts and edits 3–15 second source clips using up to six reference images, with output at 360p, 720p, 1080p, or 4K in 16:9 or 9:16.

  18. Launch

    Vercel Schedules enters public beta on all plans

    Vercel announced that Schedules is now in public beta on all plans. Schedules run tasks on a recurring cadence or at a single point in time, in production, preview, or custom environments, and can target a private Vercel Function, an HTTP path, or a Vercel Queue topic. Schedules is the successor to Cron Jobs, which continue to work, and pricing applies on Pro and Enterprise at $5 per 1M runs and $10 per 1M creations.

  19. Update

    Kimi K3.1 Pro and Pro Highspeed now available from more ZDR providers on Vercel AI Gateway

    Vercel announced that Kimi K3.1 Pro and its higher-speed variant, Kimi K3.1 Pro Highspeed, are now available on AI Gateway from more Zero Data Retention (ZDR) providers. Until this announcement, Moonshot AI was the only provider serving either model.

  20. Launch

    Vercel launches Vercel Schedules in public beta for dynamic one-time and recurring jobs

    Vercel announced Vercel Schedules, a feature that lets applications and agents create, change, or cancel one-time and recurring scheduled tasks through code without redeploying. Schedules are now in public beta and can be managed via the JS SDK, CLI, or API, with static schedules also configurable in project configuration.

  21. Launch

    Vercel Security Review enters public beta

    Vercel Security Review is now in public beta for Pro and Enterprise teams. Once enabled, it automatically reviews every push to a connected GitHub repository for vulnerabilities and reports findings with suggested fixes and AI-generated severity ratings in the Vulnerabilities dashboard.

  22. Update

    Vercel CI can now run on GitHub repositories without a linked Vercel project

    Vercel CI can now run on a GitHub repository that has no Vercel project, removing the need to create an empty project to connect the repository. Support is limited to private GitHub repositories; public repositories and other Git providers are not yet supported.

  23. Model release

    Tencent Hy4 now available on Vercel AI Gateway

    Tencent's Hy4 model is now available on Vercel AI Gateway under the model name tencent/hy4, at 50% off through December 31, 2026. Hy4 supports reasoning, tool calling, structured outputs, and image input, with a 1M-token context window.

  24. Update

    Vercel CLI adds vercel doctor command to diagnose setup problems

    Vercel CLI now includes a vercel doctor command that checks CLI version and install, login and current team, project link, and vercel.json configuration, and reports active incidents from the Vercel status page. Each check reports OK, Warning, Failed, or Skipped, along with the command to fix each problem found.

  25. Update

    Vercel adds vercel.toml support for project configuration

    Vercel announced that projects can now be configured with a vercel.toml file using TOML format. It supports the same configuration properties as vercel.json, adds comment support, and uses a flatter structure; projects should use only one of vercel.toml, vercel.json, or vercel.ts.

  26. Update

    Vercel Flags Explorer redesigned to co-locate overrides and branch recommendations

    Vercel redesigned the Flags Explorer, a tool in the Vercel Toolbar that shows active feature flags and lets users share flag overrides with teammates per branch or via URL. The redesign co-locates overrides, branch recommendations, and related actions, and teammates visiting a deployment with branch overrides set are shown a recommendation dialog.

  27. Update

    Vercel MCP now supports Client ID Metadata Documents

    Vercel's MCP server now accepts OAuth connections from any MCP client that publishes a Client ID Metadata Document (CIMD) on its own domain, instead of requiring Vercel to review and allowlist each client and its redirect URIs by hand. Users still log in and approve each connection; the consent screen shows the client's name, logo, and metadata hostname, and CIMD clients do not receive the Vercel verified badge.

  28. Update

    Vercel MCP now renders deployments as an interactive card in ChatGPT and other MCP App clients

    Vercel MCP now displays deployments as an interactive card in ChatGPT and other chat clients that support MCP Apps. The card shows live deployment status, build logs, source and commit details, and a Visit button once the deployment URL is ready.

  29. Update

    Vercel adds pre-install and during-install checks for blocked npm package versions

    Vercel now checks for blocked npm package versions before and during dependency installation in builds, to help prevent malicious install scripts from running. The blocklist contains specific package versions selected for malware or security vulnerabilities, and new rules typically reach new builds within minutes.

  30. Update

    Vercel Sandbox with Drives now starts faster

    Vercel announced that Vercel Sandboxes that mount Drives now start faster, with median creation time dropping from 1.7s to 0.7s and p95 dropping from 2.2s to 1.6s. The speedup applies automatically to all sandboxes using Drives, with no SDK upgrade or configuration change required, and pricing is unchanged.

  31. Update

    Vercel Flags adds Project Connections to evaluate flags from another project without SDK keys

    Vercel announced Project Connections for Vercel Flags, which let an application evaluate flags defined in another Vercel project within the same team without managing server SDK keys. Connected projects authenticate with their deployment identity, and for local development, vercel env pull retrieves an OIDC token.

  32. Launch

    Vercel launches Security Review in public beta

    Vercel announced Security Review, a public beta capability for Pro and Enterprise teams that scans GitHub pushes to connected repositories for vulnerabilities and reports findings as pull request comments, in the Security dashboard, and via the Vercel CLI. Vercel Agent can validate findings and, with permission, push fixes.

October 2026

  1. Update

    Vercel Data Cache is moving to Runtime Cache

    Vercel is moving Data Cache out of beta into Runtime Cache, which Next.js fetch and unstable_cache will use on every version without code changes. Existing Hobby and Pro teams can opt in early starting October 30, 2026, and Vercel plans to migrate all teams on November 29, 2026.

  2. Update

    Vercel merges Data Cache into Runtime Cache

    Vercel is consolidating its Data Cache into Runtime Cache, so Next.js fetch and unstable_cache caching now use the same storage as the existing Runtime Cache APIs. Pro and Enterprise projects receive larger storage allocations, and the Runtime Cache view gains deployment markers, revalidation causes, and write utilization.

  3. Update

    Vercel Daemons can now call services with bindings

    Vercel Daemons can now call services in their deployment through service bindings. A daemon declares a binding that names the target service and the environment variable that receives the service's URL, and the daemon calls that URL like any other HTTP endpoint while Vercel handles internal routing, authentication, and TLS and the service stays private.

  4. Model release

    Ling 3.1 Flash 1019 now available on Vercel AI Gateway for free until November 3, 2026

    Vercel announced that Ling 3.1 Flash 1019 from InclusionAI is available on AI Gateway at no cost until November 3, 2026, when requests will begin charging at the published rate. The model is designed for coding, multi-step reasoning, and tool-using agents, with a 1M-token context window.

  5. Update

    Vercel Daemons now support HTTP liveness checks

    Vercel Daemons now support configuring HTTP liveness checks. A daemon exposes an HTTP health endpoint, and its configuration accepts a livenessProbe value pointing to that endpoint; Vercel periodically requests the endpoint and restarts the daemon's container when the check fails.

  6. Update

    Vercel Connect adds webhook triggers for Resend, X, Twilio, Notion, and Telegram

    Vercel Connect now supports webhook triggers for Resend, X, Twilio, Notion, and Telegram, joining existing trigger support for Slack, Discord, Microsoft Teams, GitHub, and Linear. Connect receives provider webhook events, verifies incoming requests, handles heartbeats and verification challenges, and forwards events to an endpoint in a Vercel project.

  7. Update

    Node.js 26 LTS is now generally available on Vercel

    Vercel announced that Node.js 26 LTS is Generally Available for builds and Vercel Functions. Node.js 26 is the default for new projects, and minor and patch updates roll out automatically.

  8. Update

    Strix adds support for Vercel AI Gateway

    Strix, an open-source AI penetration testing tool, can now run models through Vercel AI Gateway, which provides access to over 400 models. Strix selects the Gateway via a STRIX_LLM setting using the vercel_ai_gateway/ prefix followed by the standard AI Gateway model ID.

  9. Pricing

    Mistral Large 4 updated weights and lower pricing available on Vercel AI Gateway

    Vercel announced that Mistral Large 4 on AI Gateway now uses updated weights, with lower prices for input, cached input, and output tokens. The existing model ID mistral/mistral-large-4 routes to the updated version without any change on the caller's side.

  10. Update

    Vercel improves IP geolocation granularity and refresh frequency

    Vercel now resolves visitor locations at finer network granularity, providing city-level data for nearly all requests and postal-code data for 90% of traffic. Geolocation data is refreshed more often and the update is enabled for all projects without code or configuration changes.

  11. Launch

    Vercel Plugin now available in Antigravity marketplace

    Vercel announced that its Vercel Plugin is now available in the Google Antigravity marketplace. The plugin gives Antigravity agents guidance on Vercel APIs and patterns, with skills covering AI SDK, Vercel Functions, Workflow, and Vercel Sandbox.

  12. Update

    Vercel Flags SDK cuts cold-start initialization time and stops usage reporting from extending function duration

    Vercel's changelog reports that @flags-sdk/vercel now loads the bundled flags datafile at client creation, removing a network wait during cold-start initialization. Usage reporting was changed to pass events to the Vercel runtime instead of using waitUntil, so it no longer extends function duration; users are told to update to version 1.5.3 or later and redeploy.

  13. Update

    Vercel Firewall adds separate controls for AI crawlers and AI assistants

    Vercel's Firewall now lets users log or block requests from AI crawlers and AI assistants separately using the AI Bots managed ruleset. The ruleset is available on all plans, and Vercel updates the rules as new bots are added.

  14. Update

    Vercel Container Registry now supports 2 GB compressed image layers

    Vercel Container Registry now supports up to 2 GB per compressed image layer, up from 500 MB, so larger layers can be pushed without splitting them. The total image size limit remains 15 GB.

  15. Model release

    Particle-1 now available on Vercel AI Gateway

    Vercel announced that Particle AI's Particle-1 model is available on AI Gateway as particle/particle-1. Particle-1 is post-trained on DeepSeek V4.1 Flash with a focus on completing software engineering tasks using fewer tokens, and has a 1M-token context window.

  16. Launch

    Paseo adds Vercel Sandbox plugin for running agents in the cloud

    Paseo now supports running fx, Codex, Claude Code, OpenCode, and Pi agents in Vercel Sandbox through a Paseo plugin. Users can start an agent in a sandbox from Paseo on a laptop and control the session from a phone, and stopping a sandbox saves files and conversation history while active processes stop.

  17. Launch

    Virtual models now available on Vercel AI Gateway

    Vercel AI Gateway now supports virtual models, which are saved configurations that specify which underlying model and settings to use. Apps and agents call a virtual model by a custom ID in the form vmc/<slug>, and its model or settings can be changed centrally without redeploying apps or reconfiguring agents.

  18. Update

    Vercel CLI now lets agents enable Web Analytics and Observability Plus without an interactive terminal

    Vercel announced that enabling Web Analytics and Observability Plus through the Vercel CLI no longer requires an interactive terminal, so agents can enable them for projects without a person running the commands. Teams must already be subscribed to Observability Plus, and the feature requires Vercel CLI version 64.2.0 or later.

  19. Launch

    Vercel CI enters beta for Turborepo projects

    Vercel announced that Vercel CI is now in beta. According to the changelog, it runs CI for Turborepo repositories with one line of configuration, and runs tasks marked with a ci tag on every push, with automatic handling of affected-only runs, Remote Cache task caching, single dependency installation per run, and dependency caching across code updates.

  20. Update

    Box Mount now runs in Vercel Sandbox

    Vercel announced that Box Mount runs in Vercel Sandbox, letting agents read and write Box files through standard filesystem operations by syncing a Box folder with a directory inside the sandbox. Box Mount is in private preview, and access is requested from Box.

  21. Launch

    Vercel AI Gateway adds 14 ElevenLabs audio models with 50% launch discount

    Vercel announced a partnership with ElevenLabs to make 14 ElevenLabs models available through Vercel AI Gateway, covering speech-to-text, streaming text-to-speech, realtime voice, music, sound effects, and multi-voice dialogue. All ElevenLabs models are offered at 50% off on AI Gateway for the first 14 days, through November 9.

  22. Launch

    Vercel adds deployment of a plain Dockerfile via the Container preset

    Vercel can now deploy a plain Dockerfile when no other framework is detected. It selects the Container preset, builds the Dockerfile, and routes all traffic to the resulting image. The Container runtime is in beta, runs on Vercel Functions, and is billed with Active CPU pricing.

  23. Update

    Vercel's fx adapter now runs the harness on the host process

    The fx adapter for the AI SDK's HarnessAgent now runs the harness and model requests in the same process as the application, using the libfx TypeScript library, while built-in filesystem and shell operations remain isolated in the sandbox. The adapter no longer installs fx inside the sandbox, runs a sandbox bridge, or exposes a sandbox port, and model credentials stay on the host instead of being forwarded into the sandbox.

  24. Launch

    Vercel Plugin now available in OpenCode

    Vercel announced that its Vercel Plugin is now available in OpenCode. The plugin provides on-demand guidance on Vercel APIs and patterns, skills for AI SDK, Vercel Functions, and Workflow, and specialist agents for deployment, performance optimization, and AI application design, plus commands to deploy projects and manage environment variables. It requires OpenCode 2 and connects to the Vercel MCP server after sign-in.

  25. Update

    Vercel adds AI traffic insights to Observability

    Vercel announced AI traffic insights in Vercel Observability, showing how AI bots access a site. The dashboard view (Observability > AI Traffic) reports requests by verified AI bot, most-crawled pages, assistant retrievals, Markdown requests, and error pages; the same metrics can be queried via the Vercel CLI, which requires Observability Plus.

  26. Pricing

    Vercel Agent sessions now cost 67% less on average

    Vercel announced that Vercel Agent sessions in the dashboard and Slack now cost 67% less on average. The reduction comes from a rebuilt context management system that uses fewer tokens per session and improves prompt-cache reuse, and no user action is required.

  27. Update

    Vercel speeds up pnpm installs by reusing dependency verification

    Vercel now reuses successful dependency verification across deployments for pnpm installs. For Pro deployments with restored build cache, Vercel observed 14% lower average install times with the optimization enabled; when dependencies are unchanged, installs can complete in milliseconds.

  28. Model release

    GPT-6.1 Terra now available on Vercel AI Gateway

    Vercel announced that OpenAI's GPT-6.1 Terra model is available through AI Gateway, using the model name openai/gpt-6.1-terra. A faster variant, openai/gpt-6.1-terra-fast, is also offered for latency-sensitive work.

  29. Update

    Vercel Connect CLI adds discovery of 1000+ services

    Vercel announced that users and agents can browse and search more than 1000 services available on Vercel Connect directly from the CLI. Results can be filtered by category, connection method, exposed product, or webhook trigger support, and the vercel connect categories command lists category slugs; both commands support JSON output. Version 64.1.0 or later of the CLI is required.

  30. Launch

    Scale AI's AgentEnv adds Vercel Sandbox provider for running agent evals

    Vercel announced that Scale AI's AgentEnv, a framework for testing agents against tasks and scoring their results, now supports running evaluations on Vercel Sandbox. The Vercel provider ships in AgentEnv 0.9.1334 and later and runs the agent and its task environment in separate Sandboxes.

  31. Launch

    Vercel adds custom monitors to Observability

    Vercel announced custom monitors in Vercel Observability, which send alerts based on query results. Each monitor tracks a single metric or a ratio of two metrics, using either statistical anomaly detection or a fixed threshold, with configurable detection window, filters, and grouping.

  32. Update

    Vercel Sandbox commands now support stdin

    Vercel Sandbox commands can now receive input through standard input (stdin). The TypeScript SDK adds stdin and detached options to runCommand(), with writeStdin() to send input and closeStdin() to end it; the Python SDK uses stdin=subprocess.PIPE with server.stdin.send() and communicate().

  33. Update

    Vercel deployments now show build cache size

    Vercel deployment summaries now display the size of the build cache each deployment saved and how close it is to the per-build-machine limit. The same data is available in the Vercel CLI version 65.1.0 or newer via vercel inspect, and through the deployment REST API endpoint.

  34. Update

    Vercel CLI adds Spend Management

    Vercel CLI now includes a vercel budgets command that lets team members view and manage the team's On-Demand Budget, showing current spend and the active budget. A vercel budgets alerts subcommand shows budget alert history, with filters for --since, --until, --limit, and --json, and a bare vercel budgets prints the three most recent alerts.

  35. Update

    Vercel MCP adds submit_feedback tool for sending feedback to Vercel

    Vercel's MCP now includes a submit_feedback tool that sends bug reports, friction reports, and feature requests to the Vercel team from an agent or harness. The tool requires sign-in and sends feedback only when the user asks the agent to share it.

  36. Update

    Vercel Connect adds managed Stripe API keys

    Vercel Connect now lets users create a Stripe connector with Managed API Key as the connection method. Vercel creates a restricted Stripe API key in the user's Stripe account, limited to selected permissions, and supports rotation from connector settings with the old key valid for up to 24 hours. The existing MCP and bring-your-own API key methods are unchanged.

  37. Update

    Vercel CLI adds feedback command for sending bugs, friction reports, and feature requests

    The Vercel CLI now includes a feedback command (vercel feedback) that sends bug reports, friction reports, and feature requests to the Vercel team from the terminal. Coding agents can also send feedback, using the -f flag or --field flag to attach structured key=value context, repeated to describe the task, expected behavior, and what went wrong.

  38. Model release

    Kimi K3.1 Pro and Kimi K3.1 Pro Highspeed available in Vercel AI Gateway

    Vercel AI Gateway now offers Moonshot AI's Kimi K3.1 Pro and Kimi K3.1 Pro Highspeed, a faster variant for interactive use. Both models keep a 1M-token context window and image input, and Moonshot AI plans to release the model weights by November 4.

  39. Pricing

    Vercel adds eight more TLDs to its free domain offer for Pro teams

    Vercel announced that eight additional top-level domains are now eligible for the free one-year domain offered to Pro teams, bringing the total eligible TLDs to sixteen. The offer is limited to one domain per team, excludes premium domains, and the domain auto-renews after the first year at the regular checkout price; the source text is partially garbled, so the full TLD list should be checked on Vercel's domains page.

  40. Launch

    Baseten's Mammoth-1 and Elephant-1 available free on Vercel AI Gateway through October 27

    Baseten's decision models Mammoth-1 (27B) and Elephant-1 (9B) are available for free on Vercel AI Gateway for one week, through October 27. Both models classify text, route requests, and score content against user-defined criteria, with a 16k-token context window and outputs that include choices, probabilities, and scores.

  41. Update

    Vercel Routing Middleware now supports Python

    Vercel announced that Routing Middleware now supports Python, for frameworks including FastAPI, Flask, and Django. Middleware runs before requests reach the app and can rewrite, redirect, or respond to them, and is configured by pointing a proxy entrypoint in vercel.toml at a middleware app.

  42. Update

    Skipped monorepo deployments no longer count against Vercel limits

    Vercel stopped counting deployments that it skips in monorepos against deployment rate limits. Vercel skips a project's deployment when its source code and internal dependencies have not changed, using Turborepo's package graph and lockfile analysis.

  43. Launch

    Vercel announces general availability of WebSocket support in Vercel Functions

    Vercel announced that WebSocket support in Vercel Functions is now generally available, following a public beta launched in late June. According to the post, Vercel CDN has served 300 million WebSocket connections since the beta and currently serves 15 million per day.

  44. Update

    DeepSeek V4.1 Flash added to Vercel Token Pass alongside Kimi K3 and GLM 5.3

    Vercel announced that Token Pass now includes DeepSeek V4.1 Flash, alongside Kimi K3 and GLM 5.3, each accessed through its preferred provider. Token Pass is a $30/month subscription for Pro teams that grants $80 in monthly credit and can be used with coding agents or with the AI SDK, with Zero Data Retention enabled by default.

  45. Update

    Vercel updates Hobby and Pro Terms of Service, effective October 19, 2026

    Vercel updated the Terms of Service for Hobby and Pro customers, effective October 19, 2026. Changes cover credit expiry and refund rules, a new section on managed accounts using company email domains, auto-renewal and cancellation terms, and statutory cancellation rights for eligible consumers in the UK and EU.

  46. Launch

    MiniMax M3.1 Flash available on Vercel AI Gateway for free until Oct 22

    MiniMax M3.1 Flash, an open-weight multimodal model for coding and agentic work with a 1M-token context window, is now available on Vercel AI Gateway. The free rate applies until Oct 22 and only when MiniMax serves the request, which requires sorting by provider in gateway order options.

  47. Model release

    Mercury 3 from Inception now available on Vercel AI Gateway

    Vercel announced that Inception's Mercury 3 model is available on AI Gateway at 50% off launch pricing. Inception reports generation speeds above 1,000 tokens per second and a 1M-token context window; Mercury 3 uses diffusion-based parallel refinement for text generation.

  48. Launch

    Vercel releases durable-harness package to make coding agent harnesses resumable after crashes

    Vercel announced durable-harness, an open source package of adapters and utilities that let existing agent harnesses (including Pi, OpenCode, Codex, and Claude Code) store their sessions in the Workflow SDK's durable World backend. It records each model and tool call intent and outcome to a durable log, and backstops turns with a queue message so a new function can resume an interrupted turn.

  49. Update

    Vercel AI Gateway logs add a sessions view showing session affinity routing

    AI Gateway logs now include a Sessions view that groups requests into sessions and shows how session affinity routed each one. Each session row lists the client app that ran it (such as Claude Code, Codex, or Cursor), along with cost, models, providers, and the share of input tokens served from cache across turns.

  50. Model release

    DeepSeek V4.1 Flash 1017 now available on Vercel AI Gateway

    Vercel announced that DeepSeek V4.1 Flash 1017, a minor update to DeepSeek V4.1 Flash, is now available on AI Gateway under the model name deepseek/deepseek-v4.1-flash-1017. According to Vercel, the update improves reasoning, coding, agentic tasks, and image understanding, and finishes tasks faster with shorter outputs than the previous snapshot.

  51. Model release

    Reflection AI's Beam now available on Vercel AI Gateway for free

    Reflection AI's Beam, a US open-weight reasoning model for coding and agentic tasks previously in stealth, is now available on Vercel AI Gateway. Beam is free to use for a limited time during the launch period when served by Reflection AI.

  52. Launch

    Vercel and Shopify release Vercel Shop, an open source storefront template for agent shoppers

    Vercel announced that Vercel Shop, an open source Next.js storefront template built with Shopify, is generally available. The template exposes product pages in machine-readable form, supports the Universal Commerce Protocol via MCP at the merchant's domain, and includes a shopping assistant that shares a cart with the storefront; the rebuilt Hydrogen SDK is also generally available.

  53. Launch

    Vercel Connect adds MCP Events support in public beta

    Vercel Connect now supports MCP Events in public beta, allowing apps and agents to subscribe to webhook events from compatible MCP servers. Connect creates and automatically renews subscriptions on the MCP server, verifies server webhooks, and forwards events to the application using authenticated delivery.

  54. Milestone

    Vercel Ship 2026 recap: Python generally available, Vercel Daemons in public beta, Agent Stack updates

    Vercel published a recap of Vercel Ship 2026, held across London, Berlin, New York, Sydney, and San Francisco. Announcements included general availability of Python on Vercel (with FastAPI, Flask, and Django support), a public beta of Vercel Daemons for continuously running background workers, and updates to the Vercel Agent Stack and Vercel for Enterprise.

  55. Launch

    Vercel Drives is now generally available

    Vercel announced that Drives, persistent storage mounted as a directory in Vercel Sandbox, is now generally available on Hobby, Pro, and Enterprise plans. Files in a Drive persist after a sandbox stops and can be mounted by later runs or other sandboxes, and Drives can be forked into independent writable copies.

  56. Update

    Vercel Connect adds Secure Compute routing for providers with IP allowlists

    Vercel Connect now supports Secure Compute. With Secure Compute routing enabled on a connector, Connect's provider requests, including token obtain and refresh requests, use the selected network's dedicated static IP addresses instead of Connect's shared IPs. Requests for initial connector creation and provider discovery bypass the selected network, and routed requests fail if the network is unavailable rather than falling back to shared IPs.

  57. Launch

    Vercel Daemons enters public beta

    Vercel announced that Vercel Daemons is now in public beta. Daemons let users run long-running containerized processes that pull and execute work from queues, databases, or third-party systems without an HTTP server, and Vercel builds, deploys, and restarts the instances.

  58. Launch

    Muse Spark Open now available on Vercel AI Gateway

    Vercel announced that Muse Spark Open, an open-weight reasoning model from Meta, is now available on AI Gateway under the model name meta/muse-spark-open. The model supports text and image input with a 1M-token context window and is accessible via OpenAI-compatible and Anthropic-compatible APIs and coding agents such as Claude Code, Codex, and Cursor.

  59. Model release

    Liquid AI's d1-ultra now available on Vercel AI Gateway

    Vercel announced that Liquid AI's d1-ultra, a multimodal decision model for classification, routing, and scoring across text and images, is now available on AI Gateway under the model name liquid/d1-ultra. The model returns structured answers with probabilities (Boolean, Choice, and Score question types) rather than generated text, supports a 32,768-token context window, and is billed on input tokens only, including image input.

  60. Launch

    Vercel launches Token Pass in early access for Pro teams

    Vercel has put Token Pass into early access for Pro teams. For $30 per month, a team receives $80 in monthly credit for GLM 5.3 and Kimi K3, with more models planned, usable through AI Gateway with coding agents or custom agents built with the AI SDK.

  61. Model release

    Vercel AI Gateway adds Cat Jam in stealth, free during stealth period

    Vercel announced that Cat Jam, a reasoning model for coding and agentic tasks with low-latency inference, is available on AI Gateway as a stealth model, free to use during the stealth period. The model supports function calling and structured outputs, accepts text input only, and supports low, medium, high, xhigh, and max reasoning efforts; Zero Data Retention is not available for it.

  62. Launch

    Vercel makes Python generally available, adds container runtime support, and launches Daemons in public beta

    Vercel announced that Python backends are now generally available, with FastAPI, Flask, and Django running on standard CPython. It also added container support for Vercel Functions on Fluid compute and introduced Vercel Daemons in public beta for long-running background workers that pull jobs.

  63. Launch

    Python on Vercel is now generally available

    Vercel announced general availability of Python on Vercel, which supports FastAPI, Flask, Django, and FastHTML with automatic framework detection, builds on Fluid compute, and provides Python SDKs for storage, queues, durable workflows, and AI models. The announcement also covers Vercel Services for running a Python backend and frontend in one project and Vercel Daemons for long-running Python workers.

  64. Update

    Vercel expands Enterprise Apps and Agents platform with governance, connectors, and AI Gateway features

    Vercel published a blog post announcing expansions to its Enterprise Apps and Agents platform, covering a single deployment target for engineers and citizen developers, central governance via existing IdPs, scoped agent connectors (Vercel Connect), and AI Gateway access to hundreds of models with Zero Data Retention. The post cites customer examples Klaviyo and Mercor and reports usage figures from the vendor.

  65. Launch

    Vercel Drives reaches general availability for Vercel Sandbox

    Vercel announced that Vercel Drives, persistent storage for agent files that is independent of the execution environment, is generally available for Vercel Sandbox on all plans. Drives can be mounted into sandboxes, forked, and shared across sandboxes, with a default size of 1 TiB (1 GiB on Hobby) and up to 16 TiB available.

  66. Launch

    Vercel makes Bring Your Own Cloud on AWS generally available

    Vercel announced that Bring Your Own Cloud (BYOC) on AWS is now generally available for Enterprise plans. BYOC lets teams run Vercel Functions, including Node.js and Python, in their own AWS account while Vercel handles deployments, environments, routing, and static assets.

  67. Milestone

    Vercel reports results of $1M Sandbox bug bounty challenge, including a Linux kernel networking vulnerability

    Vercel published findings from its $1M public challenge inviting researchers to escape Vercel Sandbox, which ran in August. Vercel reports 1,285 reports received in two weeks, $242,000 paid in bounties, and no reports showing access to customer data; the most severe finding was a Linux kernel networking defect, with fixes and CVEs pending.

  68. Launch

    Vercel AI Gateway adds free access to NVIDIA Nemotron 3.5 Lightning

    Nemotron 3.5 Lightning, an open-weight reasoning and chat model from NVIDIA, is now available for free on Vercel AI Gateway under the model ID nvidia/nemotron-3.5-lightning-free. The model has 30B total parameters with 3B active per token, a 1M token context window, and supports tool calling and structured outputs.

  69. Launch

    Amazon S3 available as an integration in the Vercel Marketplace

    Vercel announced that Amazon S3 can now be created and connected from the Vercel Marketplace within a Vercel project dashboard. The integration adds connection details as environment variables, uses short-lived OIDC credentials instead of stored AWS keys, and is also available in v0.

  70. Model release

    Drama 3 text-to-speech model now available on Vercel AI Gateway

    Vercel announced that Fish Audio's Drama 3 text-to-speech model is available on AI Gateway, currently in preview, with 80% off for 30 days through November 12. The model is called via the model name fish-audio/drama-3 and supports natural-language direction of emotion, pacing, sounds, and pauses, plus multi-speaker dialogue with a separate voice per character.

  71. Update

    Vercel speeds up repository checkout and Git LFS downloads

    Vercel is rolling out faster repository checkout and Git LFS downloads on build machines with more than two vCPUs, reducing the time spent preparing project files before a build. The average time to get project files ready fell 20%, and fell 54% for LFS-enabled projects, saving about 13 seconds per deployment on that step.

  72. Launch

    Gemini 4 Argon now available on Vercel AI Gateway

    Vercel announced that Google's Gemini 4 Argon model is available through Vercel AI Gateway under the model name google/gemini-4-argon. The model has a 1M token context window, accepts text, image, PDF, and video input, supports tool calling, and uses thinking levels (low, medium, high) instead of thinking budgets.

  73. Launch

    MAI-Image-2.6 and MAI-Image-2.6-Flash now available on Vercel AI Gateway

    Microsoft's MAI-Image-2.6 and MAI-Image-2.6-Flash image models are now available through Vercel AI Gateway. Developers can generate images from text or edit existing images using one API key without a separate provider account, via the AI SDK or AI CLI.

  74. Model release

    Muse Wonder from Meta now available on Vercel AI Gateway

    Vercel announced that Meta's Muse Wonder model (meta/muse-wonder-1.0) is now available through AI Gateway. The model has a 1M-token context window, accepts text, image, and PDF input, and supports tool calling and structured output.

September 2026

  1. Launch

    Vercel adds a path for deploying OpenAI Agents API apps

    Vercel published support on September 10, 2026 for building on the OpenAI Agents API, using Next.js, Functions, Queues and Vercel Sandbox together. OpenAI manages the agent loop and session state, while Sandbox runs the code the agent produces.

  2. Update

    Vercel Sandbox expands from four regions to twenty

    Vercel made Sandbox available in all 20 compute regions on September 10, 2026, up from four. Region selection is available on every plan; Pro and Enterprise teams can configure failover regions, which Vercel tries in order when a sandbox cannot be created in the primary. The default remains iad1 and pricing varies by region.

August 2026

  1. Launch

    Vercel adds an agent builder to its dashboard

    Vercel shipped in-dashboard agent creation on August 28, 2026. The builder scaffolds the agent's code, pushes it to a connected Git repository, and creates a Vercel project wired to AI Gateway models, a web chat or Slack front end, and MCP connections, so a deployed agent exists before any local setup.

  2. Launch

    Vercel Chat SDK adds support for Claude Managed Agents

    Vercel published a template on August 27, 2026 for running a Slack research bot on Anthropic's Claude Managed Agents through the Chat SDK. It holds one persistent Claude session per Slack thread, streams research briefs with their sources back into the thread, and needs no Slack credentials managed by the app itself.

July 2026

  1. Launch

    Vercel adds spend budgets to AI Gateway

    Vercel launched spend budgets for AI Gateway on July 31, 2026, letting teams cap spend for an entire team or a single project, with email alerts and default budgets that new projects and API keys inherit automatically.

  2. Pricing

    Vercel Sandbox drops charges for inbound data transfer

    Vercel Sandbox no longer bills for data downloaded into a sandbox — cloning repos, installing packages, or pulling external artifacts is now free. Data downloaded to a Sandbox's exposed ports is still billable, along with outbound transfer and other resource usage.

  3. Update

    GPT-5.6 and Muse Spark 1.1 land on Vercel AI Gateway

    Vercel added OpenAI's GPT-5.6 family (Sol, Luna, Terra) and Meta's Muse Spark 1.1 as selectable models on AI Gateway, its unified routing layer for calling multiple model providers through one API and billing relationship.

VercelAll release trackersFeed