DeepSeek V4.1 Flash 1017 Ultrafast model now available on Vercel AI Gateway
Developers can now call DeepSeek's latest ultrafast model through Vercel AI Gateway with unified API key and spend tracking.
Evidence
+DeepSeek V4.1 Flash 1017 Ultrafast now available on AI Gateway
Call DeepSeek V4.1 Flash 1017 Ultrafast through Vercel AI Gateway with one API key, no provider account, spend tracking, and request traces
Data Cache merging into Runtime Cache with performance and storage improvements
Developers using Data Cache will benefit from faster performance under load, 8–10x more storage capacity, and enhanced observability as the service exits beta.
Evidence
+Data Cache is exiting beta and merging into Runtime Cache, which brings faster performance under load, 8–10x more cache storage per project, and deeper observability.
Pro teams now eligible for free domains with eight additional TLDs
Vercel Pro customers can now use free domains with .co, .studio, .agency and other TLDs (eight more added), bringing the total to sixteen supported TLDs.
Evidence
+Eight more TLDs now eligible for a free domain with Pro
The free domain offer for Pro teams now covers eight more TLDs, including .co, .studio, and .agency, for sixteen in total.
Token Pass early access: $80 monthly credit for open-weight models at $30
Teams can now access featured open-weight models like GLM 5.3 and Kimi K3 at a discounted rate, enabling coding agents and personal agents with reduced costs
Evidence
+Vercel Token Pass, now in early access, gives your team $80 in monthly credit for featured open-weight models for $30, starting with GLM 5.3 and Kimi K3.
Deployment Storage billing begins for existing Pro teams
Existing Pro teams will now be charged $0.10 per GB-month for Deployment Storage and Functions Storage starting October 23, with retention window moving to 30 days.
Evidence
+Deployment Storage billing begins for existing Pro teams
Existing Pro teams will pay $0.10 per GB-month for Deployment Storage and Functions Storage, and the retention window moves to 30 days on October 23.
Skip sending request bodies to Routing Middleware to reduce Fast Origin Transfer usage
Developers can now optimize performance and costs by preventing unnecessary request body forwarding to Routing Middleware, improving Time to First Byte.
Evidence
+Skip sending request bodies to Routing Middleware
Set skipMiddlewareRequestBody to stop forwarding the client request body to Routing Middleware, reducing Fast Origin Transfer usage and helping improve Time to First Byte.
Speed Insights deprecates First Input Delay on November 1st
Users relying on First Input Delay metrics in Speed Insights will need to transition to alternative metrics, though no immediate action is required.
Evidence
−Vercel CDN no longer caches responses with Vary: Cookie
Vercel's CDN no longer stores responses whose Vary header includes Cookie. They are still served normally, but no cache entry is written. Affected requests return x-vercel-cache: MISS with the reason vary_key_denied:cookie in runtime logs.
+Speed Insights deprecates First Input Delay on November 1st
Speed Insights is deprecating First Input Delay on November 1, with no action required.
Users will see CDN Requests instead of Edge Requests in their usage metrics and documentation. Pricing and limits remain the same, but terminology alignment may affect monitoring and communication.
Evidence
−Claude Sonnet 5.5 now available on AI Gateway
Access Claude Sonnet 5.5 through Vercel AI Gateway with the AI SDK, compatible APIs, or coding agents for coding, clearer writing, and polished documents.
+Edge Requests are now called CDN Requests
CDN Requests is the new name for the usage metric previously called Edge Requests. Pricing, limits, and how usage is measured are unchanged.
Vercel Connect now supports TanStack AI MCP clients with runtime token issuance
Developers can now authenticate TanStack AI MCP clients through Vercel Connect with tokens issued at runtime and consent handled at the route boundary, reducing environment configuration complexity.
Evidence
+Vercel Connect now supports TanStack AI
Authenticate TanStack AI MCP clients with Vercel Connect. Tokens are issued at runtime and consent is handled at the route boundary, so nothing is stored in your environment.
Grok 4.7 promotional announcement removed from changelog
The 40% discount offer for Grok 4.7 on AI Gateway, eve, and fx (valid September 21) is no longer featured in the changelog, likely because the promotion period has expired.
Evidence
−Grok 4.7 now available and 40% off on AI Gateway, fx, and eve
Grok 4.7 from xAI is now available on Vercel AI Gateway, eve, and fx, and 40% off for a week starting September 21.
Hobby plan deployments retention policy changed to free up storage
Hobby plan users will have inactive deployments automatically deleted if accounts exceed 10GB storage limit.
Evidence
+Hobby projects now retain fewer deployments to free up storage
We're changing the retention exceptions for Hobby plans so that it's easier to stay within the free usage limits. Accounts above the 10GB limit will have inactive deployments automatically deleted.
Vercel Sandbox storage increased from 32 GB to 64 GB per sandbox
Developers using Vercel Sandbox will have double the default storage capacity for their sandboxed environments.
Evidence
+Vercel Sandbox now provides 64 GB of storage per sandbox, up from 32 GB, by default on the latest SDK and CLI versions and when created with an image.
Tako Search is free on AI Gateway through September 30
Developers can use Tako Search tool on AI Gateway for free until September 30th for live web search capabilities across any AI Gateway model.
Evidence
+Tako Search is free on AI Gateway through September 30
Tako Search search tool is live on AI Gateway and provides live web search for any AI Gateway model. The tool is free through September 30th.
Cursor Cloud Agents now supported on Vercel Sandbox infrastructure
Users can run Cursor Cloud Agents on Vercel's managed infrastructure with scale-to-zero workers and isolated microVMs, enabling more flexible AI agent deployment options.
Evidence
+Cursor Cloud Agents can now run in Vercel Sandbox
Run Cursor Cloud Agents on infrastructure you control with Vercel Sandbox, using scale-to-zero workers, isolated microVMs, and durable orchestration.
Claude Managed Agents now supported in Chat SDK with Slack integration
Developers can now build Slack bots using Claude Managed Agents through the Chat SDK, with persistent sessions and streamed responses without managing credentials.
Evidence
+Run Claude Managed Agents with Chat SDK
Ship a Slack research bot built on Claude Managed Agents and Chat SDK. One persistent session per thread, streamed briefs with sources, and no Slack credentials to manage.
Ling 3.0 Flash Fin available free on AI Gateway through September 25
Developers can use Ling 3.0 Flash Fin from Inclusion AI at no cost on AI Gateway until September 25, with a -free model ID that prevents billing.
Evidence
+Ling 3.0 Flash Fin now available on AI Gateway for free
Ling 3.0 Flash Fin from Inclusion AI is free on AI Gateway through September 25, with a -free model ID that stops serving instead of billing.
Algolia integrated into Vercel Marketplace for search and discovery
Vercel users can now add Algolia search capabilities directly through the marketplace without managing search infrastructure separately.
Evidence
+Algolia is now available on the Vercel Marketplace, letting Vercel teams add search and discovery to their applications without managing search infrastructure.
GPT-5.6 Sol model available at 50% discount on AI Gateway for one month
Users of Vercel AI Gateway can access GPT-5.6 Sol at reduced cost (50% off) from August 17 through September 18, lowering the barrier to using this model.
Evidence
+GPT-5.6 Sol is 50% off on AI Gateway for the next month
GPT-5.6 Sol is 50% off on Vercel AI Gateway from August 17 through September 18 on all requests running through AI Gateway.
Deploys are now up to 33% faster for apps with many ISR pages
Developers using Incremental Static Regeneration will see deployment speed improvements, with the largest gains for apps with many prerendered pages.
Evidence
+Vercel deploys are now up to 33% faster for apps with many pages that use Incremental Static Regeneration (ISR), with the biggest gains for apps that have many prerendered pages.
Regional inference now available on AI Gateway with US/EU pinning
AI Gateway users can now control request routing to specific regions (US or EU) with a single field, providing better data residency control and performance optimization across all providers.
Evidence
+Regional inference now available on AI Gateway
Pin requests to the US or EU with a single field that works across every provider, and get the serving region back on every response.
Anthropic SDK now reports thinking tokens as reasoning token usage
Developers using @ai-sdk/anthropic will see thinking tokens correctly categorized in token usage metrics, improving cost tracking and API usage visibility.
Evidence
+e29788d: fix(anthropic): report thinking tokens as reasoning token usage