AI Pulse

DeepSeek V4 Flash now runs on updated weights by default on AI Gateway, with notably stronger agentic capabilities. On Terminal-Bench, it scores 82.7, up 25.8 points from 56.9 in the April preview. Requests to deepseek/deepseek-v4-flash pick up the new weights automatically, with no change to the model ID or your code. For now, DeepSeek is the only provider serving the updated weights. Other providers, including ones with Zero Data Retention, are coming next week. To use the updated DeepSeek V4

Vercel Passport is now generally available. Passport allows you to protect your Vercel deployments with your own identity provider. Visitors authenticate through Okta, Microsoft Entra ID, or any OIDC provider before viewing a protected deployment, and Vercel forwards a signed identity token to the deployment so application code can build on who the visitor is. Read visitor identity in application code The getIdentity() helper in @vercel/passport reads the Vercel request context and returns the a

<h3>Misc Changes</h3> <ul> <li>[turbopack] Respect <code>sideEffects</code> in a <code>package.json</code> file with <code>optimizePackageImports</code>: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5018084051" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96383" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96383/hovercard" href="https://github.com/vercel/next.js/pull/96383">#96383<

<h2><a href="https://github.com/googleapis/python-genai/compare/v2.15.0...v2.16.0">2.16.0</a> (2026-07-29)</h2> <h3>Features</h3> <ul> <li>Add environment resource (<a href="https://github.com/googleapis/python-genai/commit/615b6c3795f3ac005a7ed9517ddc2c86b2b7a043">615b6c3</a>)</li> <li>Support GoogleMaps Tool grounding_types places and routing (<a href="https://github.com/googleapis/python-genai/commit/95a335d809d75b987303d5a52e533a938585b7b5">95a335d</a>)</li> <li>Wire environment service into

Vercel's CDN will begin passing through the Server-Timing response header to the client on August 10, 2026. Use Server-Timing to report backend metrics like database query time and cache hits. These values appear in the browser's network panel and as PerformanceServerTiming entries through the Performance API. If you want to keep the current behavior, where the header is stripped from every response, you can add a transform in vercel.json : Learn more about the Performance API and Server Timing

The @vercel/sandbox SDK now supports multiple Linux users and groups, so you can run agents side by side in a single Sandbox. Each agent runs as its own user with a private home directory. A group opens a shared workspace when they need to collaborate. This makes multi-agent systems easier to build. Call createUser for each agent; its commands and file operations run as that user, and users can't read, write, or list each other's files. To set up a shared directory, call createGroup and add user

Shopify and Vercel Open source and runtime agnostic, runs anywhere JavaScript does Standard Actions brings agentic commerce to every storefront Feature development cut from months to a week for retailers like Global Retail Brands Shopify powers commerce for millions of merchants worldwide, and Hydrogen is its framework for building headless storefronts. The Shopify team is partnering with Vercel to rebuild Hydrogen from the ground up. The new version is open source and runtime agnostic so develo

MiniMax H3 is now available on AI Gateway. H3 generates 2K video from a text prompt, a starting image, a pair of first and last frames, or reference material. Alongside text-to-video and first-frame image-to-video, the model supports first-to-last keyframe transitions and multimodal reference-to-video, conditioning a generation on reference images, video, or audio in a single request. Reference and keyframe modes are mutually exclusive. Output is mp4 at 2K resolution, from 5 to 15 seconds, in as

You can now exchange OIDC tokens from CI/CD workflows, including GitHub workflows , for short-lived Turborepo access tokens. These tokens grant access to Vercel's Remote Cache , and are a more secure alternative to long-lived Personal Access Tokens (PATs). OIDC tokens are short-lived, only grant access to Vercel Remote Cache, and are associated with your Vercel team, rather than a specific team member. We recommend all customers migrate their CI/CD workflows from PATs to OIDC. Get started by add

Inkling Small from Thinking Machines is now available on AI Gateway. Inkling Small reaches performance comparable to the larger Inkling model at about a quarter of the size, using much less compute per task. It is a broad generalist with native reasoning over audio and images, and it holds up well on reasoning, agentic coding, and tool use. Controllable thinking effort lets you trade quality against cost and latency, from minimal to maximum reasoning. For visual tasks, it can crop, zoom, and ins

You can now create Vercel Access Tokens that are limited to a project to authenticate and use the Vercel API . A project-scoped token can only read and write resources belonging to a project that the token is scoped to. Requests to any other project, a user-level resource, or a team-level resource will be denied. This ensures jobs, tools, or workflows only ever access the projects they are scoped to. Creating a project-scoped token Navigate to the Account Tokens page , found under the Settings a

Enterprise customers can now apply a portion of their Flexible Commitment toward eligible resources purchased through the Vercel Marketplace. With Flex Commit support, eligible Marketplace cost can draw directly from your existing commitment making it easier to provision the infrastructure and services your applications need. We're rolling this out with an initial set of Marketplace partners, including Neon , Supabase , and Redis , with support for more providers coming. Eligible integrations ar

mcp-handler@2.0.0 is now available with support for the 2026-07-28 Model Context Protocol specification and MCP TypeScript SDK v2 . We originally built mcp-handler to make it easier to spin up MCP servers on popular web frameworks including Next, Nuxt, Svelte, and more. With the new 2.0 release, the handler now supports: The stateless 2026-07-28 protocol, served natively, including per-request metadata and server/discover A stateless compatibility layer for clients using 2025-era Streamable HTTP

Over the past few months, we've made deployments up to 7 seconds faster end to end. We removed up to 5 seconds of fixed platform overhead from every build, and deploys from the latest Vercel CLI save up to 2 seconds more. The improvements are most noticeable on smaller builds, where orchestration represents a larger share of the total build time. The largest platform gains: ~2.2s : Internal build-process shutdown now happens outside the critical path to deployment readiness ~930ms : The Vercel C

On AI Gateway , GPT-5.6 Luna and GPT-5.6 Terra are now cheaper and GPT-5.6 Sol is faster. AI Gateway adds no markup on token pricing, so these changes reach you at the upstream rate. The changes apply to both short and long context pricing. Model Change Input: Short context (per 1M tokens) Output: Short context (per 1M tokens) openai/gpt-5.6-luna 80% price reduction $0.2 $1.2 openai/gpt-5.6-terra 20% price reduction $2 $12 openai/gpt-5.6-sol Same price, fast mode now 2.5x faster (up from 1.5x) U

<p>v16.3.0-canary.103</p>

<h2><a href="https://github.com/googleapis/python-genai/compare/v2.14.0...v2.15.0">2.15.0</a> (2026-07-28)</h2> <h3>Features</h3> <ul> <li>[GenerateContentConfig] Add GenerationConfig.audio_transcription_config and Part.audio_transcription. (<a href="https://github.com/googleapis/python-genai/commit/0f775f11ee7fd84433dd16252fe37698beebe296">0f775f1</a>)</li> <li>Add flat <code>language_codes</code> field to <code>AudioTranscriptionConfig</code>. (<a href="https://github.com/googleapis/python-gen

Grok Voice Think Fast 2.0 from xAI is now available on AI Gateway. It is a speech-to-speech voice model that takes audio in and audio out, improving on the previous Grok Voice model in reasoning, transcription accuracy, and conversation. The model reasons in parallel with speech, so it can think through a query while talking without adding latency. It has also been trained to use fewer reasoning tokens than before, so tool calls fire sooner, often before the end of the agent's first sentence. Tr

AI Gateway has a unified fast mode abstraction, now in beta. You can now request fast mode the same way for every model on AI Gateway. Set speed to fast , and the gateway serves the fast tier when it's available and falls back to standard speed when it isn't. Fast mode trades a higher per-token cost for lower latency or higher throughput. You get it wherever it is available, without pinning a provider or wiring anything up yourself. Requesting fast mode Set speed to fast under providerOptions.ga

Edge Config is now Global Config . This rename better reflects that it is a globally replicated data store with ~1ms reads in every region, built for the configuration that applications read at runtime, such as feature flags, redirects, and experimentation settings. Global Config stores can now hold up to 1 MB on every plan, and Pro and Enterprise teams can create unlimited stores with more writes per day. The rename also ships with a new server-side SDK, @vercel/global-config , and a new enviro

You can now discover and install integrations for eve agents directly from the eve CLI. Integrations come from the official eve catalog and third-party sources. Run eve add from your eve project to install an integration: Integrations write their files directly into your project and can add anything an eve agent uses, from a single tool to a channel to a full extension. Review the generated files and add any required configuration before running your agent. Find integrations with the new eve reg

Pro and Enterprise teams can now purchase additional custom environment capacity without contacting sales. Custom environments let you model your team's release process on Vercel. You can add staging , qa , or any named stage between preview and production , and each environment gets its own branch tracking, environment variables, and domains. You can purchase or adjust capacity from the dashboard, API, or CLI. Dashboard Project → Settings → Environments → Custom Environments API POST /v1/projec

Sign in with ChatGPT adds your ChatGPT account as an authentication option for Vercel. It is available when adding the Vercel plugin to ChatGPT and when signing in to Vercel or v0. When you add the Vercel plugin to ChatGPT, you can easily sign in to Vercel and grant team and project permissions without leaving ChatGPT. "Continue with ChatGPT" also appears as a sign-in option on the Vercel and v0 sign-in and sign-up pages. Team requirements like 2FA and SSO still apply. You can re-authenticate or

<h3>Misc Changes</h3> <ul> <li>Revert "Revert "[turbopack] Track re-exports in <code>import_usage</code> inside of <code>compute_import_usage</code>"": <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5003125539" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96315" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96315/hovercard" href="https://github.com/vercel/next.js/pull/96315">#96315</a

<h3>Misc Changes</h3> <ul> <li>Turbopack server hmr: avoid complete <code>clear</code> on graph changes: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4825175795" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/95546" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/95546/hovercard" href="https://github.com/vercel/next.js/pull/95546">#95546</a></li> <li>Turbopack HMR: extract deferring bui

<h3>Misc Changes</h3> <ul> <li>Improve root detection to handle worktrees, more workspaces &amp; stray lockfiles: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4971362829" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96159" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96159/hovercard" href="https://github.com/vercel/next.js/pull/96159">#96159</a></li> <li>Revert "[turbopack] Track r

<h3>Misc Changes</h3> <ul> <li>Unify allow-runtime with Partial Prefetching: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4962358904" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96106" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96106/hovercard" href="https://github.com/vercel/next.js/pull/96106">#96106</a></li> <li>Attempt static prefetch before resorting to runtime: <a class="i

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/unnamed_2_vNnOv20.max-600x600.format-webp.webp">We’re announcing even more new capabilities in Managed Agents in Gemini API so developers can build reliable, production-ready agents.

<p>Initial release</p>

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Dinner_parties.max-600x600.format-webp.webp">These AI features can help you craft a menu, design a tablescape, and handle other party-planning tasks.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/AI_Mode_real_world.max-600x600.format-webp.webp">It might sound counterintuitive, but Search's AI tools can actually help you make the most of your time offline whether you want to book concert tickets or find the perf…

Vercel Sandbox now supports forking with Sandbox.fork() . The fork starts from the source's current snapshot and inherits its config and environment variables. If the source is running, it forks the latest saved state, not the live in-memory state. If the source has no snapshot, it falls back to a fresh create, using the source's runtime and config. Any parameter you pass overrides the inherited value. A fork takes about the same time as creating a sandbox, with the same limits. Use it to branch

Vercel Connect now lets you link a connector to a Custom Environment so deployments there can request provider tokens and receive forwarded webhooks. Custom Environments appear alongside Production, Preview, and Development when you add or edit a project on a connector in the dashboard. From the CLI, pass the environment's slug to --environment : Passing --environment replaces the default environments. Once linked, a deployment in that environment can request tokens with getToken as usual. If th

<p>v16.3.0-canary.98</p>

AI Gateway now supports regional inference . Set inferenceRegion on a request to pin it to the US or EU. Every model provider that supports the selected region handles it the same way. Inference runs there, and any data the provider keeps is stored there. AI Gateway supports two pinned regions, plus global routing: Region Where inference runs us A US data center eu An EU data center global Any region If no model provider can serve it, the request fails rather than running somewhere else. Every r

eve agents on Slack can now keep replying in a thread without repeated mentions, cancel an in-progress response or reset a conversation entirely, and react to any event your Slack app subscribes to. Continue conversations without repeated mentions Mentions no longer have to carry the conversation. Once a thread has an active session, your agent can reply on its own. The new onMessage hook receives incoming Slack messages, and two helpers decide which ones to handle: ctx.isBotMentioned() detects

Sandstone on Vercel 1,000+ legal requests managed daily across customer teams 7-app turborepo monorepo deployed seamlessly to multiple Vercel projects Multi-step agentic legal workflows built end-to-end with AI SDK When in-house legal teams get a request from their business, it kicks off a manual process of pulling data from multiple systems. In a typical workflow, lawyers will pull deal details from Salesforce, review past contracts, and chase account teams for missing context, all before they

Last week, OpenAI evaluated two models on an exploit benchmark within an isolated sandbox. Guardrails were reduced for testing, and the models found a vulnerability in their environment, accessed the internet, and reached Hugging Face's production database. No human directed the action, but the breach is a clear example of how much more capable malicious attackers are when equipped with powerful AI models. But defenders have the same tools, and a clear advantage: knowledge of their own codebase.

AI Gateway now supports WebSocket mode for the OpenAI Responses API. You can keep a persistent connection open and continue each turn by sending only new input items plus previous_response_id , instead of re-sending the full context over a fresh HTTP request every turn. OpenAI reports up to ~40% faster end-to-end execution on WebSockets for agentic rollouts with 20 or more tool calls. Route Responses API traffic over WebSocket The Responses route opens at GET /v1/responses and accepts raw respon

The Nuxt team has released Nuxt 4.5.1 and 3.21.10, along with @nuxt/devtools 3.3.1, to address eight security advisories, including a high-severity server-side remote code execution vulnerability. Vulnerabilities addressed Vulnerability Severity Advisory Server-side remote code execution via server island props High GHSA-9473-5f9j-94wq Unauthorized component instantiation via server island props Medium GHSA-48hr-524c-v5w3 Route rule authorization bypass High GHSA-hxvh-4h3w-prp9 Server component

You can now run Claude Managed Agents with Chat SDK . Claude Managed Agents handles the agent loop server-side, including the model, tools, session state, and sandboxed web research. Chat SDK gives that agent a chat interface through a single type-safe handler, with adapters that carry it to Slack, WhatsApp, and more. What you get Token-by-token streaming : Replies render as the model writes them, over a single streamed response. Live activity feed : Tool calls and model requests are available a

Kimi K3 from Moonshot AI and its faster serving path, Kimi K3 Fast , are now available from US-based providers on AI Gateway, including Baseten and Fireworks. Zero Data Retention (ZDR) is also supported for both models. Running Kimi K3 on US-based providers lets teams with data residency and compliance requirements use the model on US infrastructure. Because AI Gateway serves the models from multiple providers, it automatically routes across them for failover, higher uptime, and more available t

<h3>Misc Changes</h3> <ul> <li>fix(sandbox): release one-shot timeout ids after they run: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4971395147" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96161" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96161/hovercard" href="https://github.com/vercel/next.js/pull/96161">#96161</a></li> <li>Fix dev overlay symbolication for project paths nee

<h2>What's Changed</h2> <ul> <li>Backport/docs fixes 16.2 - July round by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/icyJoseph/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/icyJoseph">@icyJoseph</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4943744779" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issu

<h2>What's Changed</h2> <ul> <li>[15.5] Reject TypeScript >= 7.0 with an actionable error by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/lukesandberg/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/lukesandberg">@lukesandberg</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4962968982" data-permission-text="Title is private" data-url="https://gi

<h2>What's changed</h2> <ul> <li>Bug fixes and reliability improvements</li> </ul>

<h3>Misc Changes</h3> <ul> <li>[sourcemaps] Use file: sourcemaps for Turbopack to improve dev performance: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4925171960" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/95946" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/95946/hovercard" href="https://github.com/vercel/next.js/pull/95946">#95946</a></li> <li>Give RouteCacheEntry a single hidd

The Vercel WAF can now protect a Vercel Blob store. The same rules that guard your deployments (deny, challenge, rate limit) now apply to blob traffic with no changes to your code, blob URLs, or @vercel/blob . Every blob is already served through Vercel's CDN , so protection is a switch on the store, not a new proxy. Stop scrapers, geo-restrict downloads, rate limit expensive assets, or block abusive IPs before a byte is served. Rules evaluate at the edge, matching on IP, country, path, and more

<h2>What's changed</h2> <ul> <li>Added Claude Opus 5 (<code>claude-opus-5</code>), now the default Opus model — 1M context, fast mode at $10/$50 per Mtok</li> <li>Added <code>sandbox.network.strictAllowlist</code> setting to deny non-allowlisted hosts for sandboxed commands without prompting</li> <li>Added <code>DirectoryAdded</code> hook that fires after <code>/add-dir</code> or the SDK <code>register_repo_root</code> control request registers a new working directory mid-session</li> <li>Added

<h2>0.115.0 (2026-07-24)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.114.0...sdk-v0.115.0">sdk-v0.114.0...sdk-v0.115.0</a></p> <h3>Features</h3> <ul> <li><strong>api:</strong> add claude-opus-5 model (<a href="https://github.com/anthropics/anthropic-sdk-typescript/commit/cdd3606e92745b8a6af07819469cf5dc0636edb2">cdd3606</a>)</li> <li><strong>api:</strong> add tool addition/removal blocks and tool_change events (<a href="https://github.co

Workflow steps on Pro and Enterprise plans can now run for up to 30 minutes (1800 seconds), up from 800 seconds, using extended function durations (in beta). To opt in, set VERCEL_ENABLE_WORKFLOW_EXTENDED_MAX_DURATION to 1 in your project's Environment Variables, then redeploy. Requires Fluid compute and a supported Node.js or Python runtime. Extended durations are not available on Hobby plans, which remain limited to 5 minutes (300 seconds). See function duration limits by plan in the documenta

Claude Opus 5 from Anthropic is now available on AI Gateway. Opus 5 improves on previous Opus models for long-horizon agentic coding, handling multi-file features, larger refactors, and end-to-end feature work, and completing full tasks rather than leaving stubs or placeholders. Opus 5 is effective at low and medium effort, which produce quality at a fraction of the tokens and latency of higher settings. Vision is stronger on charts, documents, diagrams, and UI replication, and Opus 5 makes effe

<h3>Misc Changes</h3> <ul> <li>Turbopack: Very minor improvements for watcher loop: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4961586935" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96103" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96103/hovercard" href="https://github.com/vercel/next.js/pull/96103">#96103</a></li> <li>Turbopack: Refactor watcher event handling and batching l

<h2>0.114.0 (2026-07-23)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.113.0...sdk-v0.114.0">sdk-v0.113.0...sdk-v0.114.0</a></p> <h3>Features</h3> <ul> <li><strong>api:</strong> add new stop reason 'model_context_window_exceeded' (<a href="https://github.com/anthropics/anthropic-sdk-typescript/commit/1ec71c165d9737cb317e3f8fed818f7a1b8169ad">1ec71c1</a>)</li> </ul>

<h3>Misc Changes</h3> <ul> <li>[Bench] Add client-trace attribution pass and document metrics to render-pipeline: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4895322088" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/95828" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/95828/hovercard" href="https://github.com/vercel/next.js/pull/95828">#95828</a></li> <li>Turbopack: Split up turbo-t

Vercel now supports WebSocket connections for Python applications. WebSockets enable bidirectional communication between client- and server-side code, powering real-time features like interactive AI streaming, real-time chat, and multiplayer live collaboration. Both ASGI and WSGI applications are supported, including frameworks like FastAPI, Django, and Flask. Explore the FastAPI AI Chat and Flask AI Chat examples, or read the WebSockets documentation to get started. Read more

You can now add GitHub tools to your eve agent as an extension . Add the package, drop one file in agent/extensions/ , and your agent gets all 42 tools with Vercel Connect auth, presets, and approval rules built in. Install @github-tools/eve-extension : Then register it from a file in agent/extensions/ : Connector-backed auth: Pass a Vercel Connect connector and the extension mints short-lived, scoped GitHub tokens at runtime. Presets scope the toolset: code-review , issue-triage , repo-explorer

Vercel Flags version history can now be inspected from the Vercel CLI with the new vercel flags versions command. Run vercel flags versions <flag> to print the full revision history for a flag, with each revision's author, message, timestamp, and changed environments. Filter to a specific environment with --environment , paginate with --limit and --cursor , or add --json for scripting. To compare a revision against the one before it, run vercel flags versions diff <flag> --revision <n> . The dif

You can now connect to any running Sandbox from the Vercel dashboard. Run commands, browse the filesystem, upload and download files, and inspect open ports without leaving the browser. From the same view, you can also manage the sandbox lifecycle: take snapshots, and stop or resume persistent sandboxes. Open the Connect tab on any sandbox to try it out. Learn more about Vercel Sandboxes in the documentation . Read more

Vercel Flags now shows a live evaluation view on each flag's detail page. You can see evaluations per minute charted over time, with each flag version marked in the chart so you can tie evaluation shifts to specific configuration changes. You can group and filter evaluations by variant, reason, environment, SDK key, client, and reporting project. Custom clients can set the new clientName property on the client initializer to show up as their own group. This makes it straightforward to confirm ro

The Vercel MCP server can now deploy code directly to a new or existing project. When your AI assistant finishes building something, it can ship it to Vercel and hand back a shareable URL without leaving the chat. Point the deploy_to_vercel tool at your files and Vercel creates the project, detects the framework, installs dependencies, and builds. You get a URL you can open and share while the build finishes in the background. To get started, connect the Vercel MCP server to Claude, Cursor, or a

Ling 3.0 Flash from Ant Group is now available on AI Gateway. The model is free to use for the next three weeks, through August 3rd. Ling 3.0 Flash is a Mixture-of-Experts model with 124B total parameters and about 5.1B active per token. It has a 256K token context window and runs in thinking and non-thinking modes. Ling 3.0 Flash is built for token-efficient agentic inference at production scale, doing more work within tighter token, latency, and cost budgets across multi-step agent runs. The m

<p>v16.3.0-canary.94</p>

<h2>What's changed</h2> <ul> <li>Changed <code>/code-review</code> to run as a background subagent, so review work no longer fills your conversation and keeps stacked slash commands as its review target</li> <li>Added screen-reader announcements of deleted text for word and line deletions (<code>Option+Delete</code>, <code>Ctrl+W</code>, <code>Cmd+Backspace</code>, <code>Ctrl+U</code>, <code>Ctrl+K</code>) in <code>--ax-screen-reader</code> mode</li> <li>Fixed Windows paths with <code>\u</code>-

<h2><a href="https://github.com/googleapis/python-genai/compare/v2.13.0...v2.14.0">2.14.0</a> (2026-07-22)</h2> <h3>Features</h3> <ul> <li>[GenerateContentConfig] Add GenerationConfig.audio_transcription_config and Part.audio_transcription. (<a href="https://github.com/googleapis/python-genai/commit/dc3d78dc228709c0968554694a884f9fe17a36bc">dc3d78d</a>)</li> </ul> <h3>Bug Fixes</h3> <ul> <li>Add deprecation warnings to Imagen generate_images, edit_images, generate_videos (if using prompt/text/im

<h2>0.113.0 (2026-07-22)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.112.5...sdk-v0.113.0">sdk-v0.112.5...sdk-v0.113.0</a></p> <h3>Features</h3> <ul> <li><strong>api:</strong> add support for Managed Agents model effort, initial session events, and threads delta streaming (<a href="https://github.com/anthropics/anthropic-sdk-typescript/commit/83fef1e5a65f020750b1f90c46d748a415b6f075">83fef1e</a>)</li> </ul>

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Unpacked_hero.max-600x600.format-webp.webp">We shared how Samsung users can boost productivity and get time back on new foldables, watches, and glasses coming soon.

<h3>Misc Changes</h3> <ul> <li>Always consult <code>npm_config_user_agent</code> first: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4906366847" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/95879" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/95879/hovercard" href="https://github.com/vercel/next.js/pull/95879">#95879</a></li> <li>Rewrite next-cache-components-optimizer around a test

You can now package tools, connections, skills, instructions, and hooks into extensions that any eve agent can import. Extensions can be published to package registries like npm, then installed, versioned, and upgraded like any other project dependency. A browser-use extension might ship tools for navigating a site, a memory extension can capture context with hooks and recall it with tools, and a self-improvement extension pairs hooks with dynamic instructions. Scaffold a new extension with a si

AI Gateway now supports streaming transcription . Previously, transcription required a complete audio file and returned the full transcript in a single response. Now you can stream audio in as it's captured and get transcript updates back as the model produces them, keeping latency low for uses like live captioning and voice input. Streaming transcription is in beta and available through the AI SDK 's streamTranscribe function with any streaming-capable transcription model. The example below str

<h3>Misc Changes</h3> <ul> <li>docs: attribute App Shell prefetch to Partial Prefetching: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4940268426" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/96003" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/96003/hovercard" href="https://github.com/vercel/next.js/pull/96003">#96003</a></li> <li>Turbopack: Remove chunk group id from value of chun

<h2>0.112.5 (2026-07-21)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.112.4...sdk-v0.112.5">sdk-v0.112.4...sdk-v0.112.5</a></p> <h3>Chores</h3> <ul> <li><strong>api:</strong> add support for new refusal category (<a href="https://github.com/anthropics/anthropic-sdk-typescript/commit/479efe8ade0078966e28d754fa3085c75b6d2a58">479efe8</a>)</li> <li><strong>internal:</strong> codegen related update (<a href="https://github.com/anthropics/anth

<h2>What's changed</h2> <ul> <li>Added emoji shortcode autocomplete in the prompt input: type <code>:heart:</code> to insert ❤️, or <code>:hea</code> for suggestions — disable with the <code>emojiCompletionEnabled</code> setting</li> <li>Added warnings when transcript writes are failing (e.g. disk full) or when session saving is off due to an inherited environment variable, instead of losing transcripts silently</li> <li>Fixed a memory leak where truncated MCP tool outputs kept the full untrunca

<h3>Example Changes</h3> <ul> <li>fix: error handling and loading states in with-apollo-and-redux example: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4082575538" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/91457" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/91457/hovercard" href="https://github.com/vercel/next.js/pull/91457">#91457</a></li> </ul> <h3>Misc Changes</h3> <ul> <li>U

<h3>Example Changes</h3> <ul> <li>fix: error handling and loading states in with-apollo-and-redux example: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4082575538" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/91457" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/91457/hovercard" href="https://github.com/vercel/next.js/pull/91457">#91457</a></li> </ul> <h3>Misc Changes</h3> <ul> <li>U

<p>This release contains security fixes for the following advisories:</p> <p>High:</p> <ul> <li><a href="https://github.com/vercel/next.js/security/advisories/GHSA-m99w-x7hq-7vfj">Denial of Service in App Router using Server Actions</a></li> <li><a href="https://github.com/vercel/next.js/security/advisories/GHSA-6gpp-xcg3-4w24">Middleware / Proxy bypass in App Router applications using Turbopack and single locale</a></li> <li><a href="https://github.com/vercel/next.js/security/advisories/GHSA-p9

<p>This release contains security fixes for the following advisories:</p> <p>High:</p> <ul> <li><a href="https://github.com/vercel/next.js/security/advisories/GHSA-m99w-x7hq-7vfj">Denial of Service in App Router using Server Actions</a></li> <li><a href="https://github.com/vercel/next.js/security/advisories/GHSA-6gpp-xcg3-4w24">Middleware / Proxy bypass in App Router applications using Turbopack and single locale</a></li> <li><a href="https://github.com/vercel/next.js/security/advisories/GHSA-p9

<h2><a href="https://github.com/googleapis/python-genai/compare/v2.12.1...v2.13.0">2.13.0</a> (2026-07-21)</h2> <h3>Features</h3> <ul> <li>A new field <code>custom_vocabulary</code> is added to message <code>.google.cloud.aiplatform.v1beta1.BidiGenerateContentSetup</code> (<a href="https://github.com/googleapis/python-genai/commit/4eeb1ade8dedb9ccdabe00892d4ff2119b54a444">4eeb1ad</a>)</li> <li>Add model selector (<a href="https://github.com/googleapis/python-genai/commit/bf3dba460092d2f3af7a2b29

Searchable on Vercel 5x increase in development velocity 100+ billion tokens processed Customer-requested features shipped in as little as 30 minutes Zero model SDK implementation or API key rotation with AI Gateway Searchable helps brands track and improve how they appear across AI search engines like ChatGPT, Perplexity, and Claude, pairing visibility analytics with an agent that guides users on what to do next. The Searchable team builds on Vercel's AI SDK and AI Gateway to test new models wi

AI Gateway now supports service tiering. Service tiers let you optimize for latency, throughput, and cost per request to match your use case. Pick a faster tier for interactive workloads (less queueing, higher token throughput), or a lower cost tier for background jobs that can tolerate more latency. At launch, service tiering is available for OpenAI and Gemini models. Service tiers work across every AI Gateway API format: AI SDK , Chat Completions API , Anthropic Messages API , OpenAI Responses

Vercel Connect now includes preset connectors for 90+ services, including Shopify, Okta, Workday, Jira, and Sanity. Preset connectors are predefined configurations for supported services. They reduce manual setup by pre-populating the brand name, icon, auth type, and MCP or discovery URL. Unlike managed connectors, preset connectors don't register your app with the external service for you. Add a connector Select a preset from the connectors directory in the Vercel dashboard. Review the pre-popu

Vercel now compiles Python functions to bytecode at build time. In our benchmarks, cold starts for the median-sized function dropped from 2.8s to 1.3s . When Python imports a module without cached bytecode, it parses and compiles the source before executing it. That compilation step adds startup time for functions with large dependency trees. Vercel now compiles both application code and dependencies and includes the resulting .pyc files in the function bundle, so the interpreter skips compilati

Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are now available on AI Gateway. Gemini 3.6 Flash improves quality across coding, agentic tasks, and web development while consuming fewer tokens and making fewer model calls. It produces cleaner web and app development output. Gemini 3.5 Flash Lite upgrades the agentic capabilities of the Flash-Lite tier, making it a good fit for subagents that handle scoped parts of a larger task. To use them, set model to google/gemini-3.6-flash or google/gemini-3.5-

Laguna S 2.1 from Poolside is now available on AI Gateway. There are 2 versions of the model available: Free version (256K context window): poolside/laguna-s-2.1-free Paid version (1M context window): poolside/laguna-s-2.1 Laguna S 2.1 is an open-weight Mixture-of-Experts model that supports a context window of up to 1M tokens and runs in thinking and no-thinking modes. The model specializes in agentic coding and long-running tasks, including writing and debugging code, running tests, building b

Vercel MCP now supports purchasing Vercel products. You can: Upgrade your team to the Pro plan Add prepaid credits for v0 (requires a paid v0 plan) or AI Gateway Purchase the SIEM add-on (requires an Enterprise plan) Purchase and register a domain Vercel MCP quotes the price, explains whether the charge is one-time or recurring, and completes the purchase only after you confirm. When an exact price isn't available, it links to the relevant pricing or billing page. Purchases require a team role w

<h3>Misc Changes</h3> <ul> <li>[turbopack] Skip redundant top-level root updates: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4913406830" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/95903" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/95903/hovercard" href="https://github.com/vercel/next.js/pull/95903">#95903</a></li> <li>[turbopack] Drop unused exports from a CJS module: <a class

<h2>What's changed</h2> <ul> <li>Added <code>sandbox.filesystem.disabled</code> setting to skip filesystem isolation while keeping network egress control</li> <li>Fixed a slowdown in long sessions where message normalization cost grew quadratically with the number of turns, causing multi-second stalls and slow resumes</li> <li>Fixed auto mode denying commands with "HTTP 401" classifier errors after the OAuth token expired or rotated mid-session</li> <li>Fixed AskUserQuestion telling Claude to co

<h2>0.6.1 (2026-07-20)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/aws-sdk-v0.6.0...aws-sdk-v0.6.1">aws-sdk-v0.6.0...aws-sdk-v0.6.1</a></p> <h3>Bug Fixes</h3> <ul> <li><strong>aws:</strong> preserve AWS options and auth mode across withOptions() (<a href="https://github.com/anthropics/anthropic-sdk-typescript/issues/214" data-hovercard-type="pull_request" data-hovercard-url="/anthropics/anthropic-sdk-typescript/pull/214/hovercard">#214</a>) (<a

<h2>0.112.4 (2026-07-20)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.112.3...sdk-v0.112.4">sdk-v0.112.3...sdk-v0.112.4</a></p> <h3>Bug Fixes</h3> <ul> <li><strong>aws:</strong> preserve AWS options and auth mode across withOptions() (<a href="https://github.com/anthropics/anthropic-sdk-typescript/issues/214" data-hovercard-type="pull_request" data-hovercard-url="/anthropics/anthropic-sdk-typescript/pull/214/hovercard">#214</a>) (<a href=

Team Owners can now clear the team's Remote Cache of all artifacts in one click. This is useful when you believe there are poisoned artifacts in your cache. In your team's Build and Deployment settings, visit the Remote Caching section and clear the Remote Cache. Visit the docs to learn more. Read more

Vercel Workflows now keeps each run's state, queue dispatch, and output streams in a single home region: the region where the run starts by default, or any target region you choose. A run keeps its home region for its lifetime, so for agents built on Workflows, the whole loop stays near the user: an agent serving someone in Sydney executes, checkpoints its progress, and streams output from Sydney. During a regional incident, workflow traffic fails over to the next closest region. To get started,

<p>v16.3.0-canary.90</p>

<h2>What's changed</h2> <ul> <li>Claude no longer runs the <code>/verify</code> and <code>/code-review</code> skills on its own; invoke them with <code>/verify</code> or <code>/code-review</code> when you want them</li> </ul>

<h2>What's changed</h2> <ul> <li>Fixed single-segment <code>dir/**</code> allow rules like <code>Edit(src/**)</code> auto-approving writes to nested <code>dir/</code> directories anywhere in the tree instead of only <code><cwd>/dir</code></li> <li>Fixed a permission-check bypass affecting commands run in Windows PowerShell 5.1 sessions</li> <li>Fixed Bash permission checks to fail closed on file-descriptor redirect forms that bash parses differently than the permission analyzer</li> <li>Fixed Ba

<h3>Misc Changes</h3> <ul> <li>Fix Rust doctest failures: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4915113066" data-permission-text="Title is private" data-url="https://github.com/vercel/next.js/issues/95909" data-hovercard-type="pull_request" data-hovercard-url="/vercel/next.js/pull/95909/hovercard" href="https://github.com/vercel/next.js/pull/95909">#95909</a></li> <li>[turbopack] Clean up <code>server_chunking_context</code>: <a class="issue-link js-

<h2>0.0.6 (2026-07-17)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/google-cloud-sdk-v0.0.5...google-cloud-sdk-v0.0.6">google-cloud-sdk-v0.0.5...google-cloud-sdk-v0.0.6</a></p> <h3>Bug Fixes</h3> <ul> <li><strong>google-cloud:</strong> bump google-auth-library to ^10.2.0 (<a href="https://github.com/anthropics/anthropic-sdk-typescript/issues/230" data-hovercard-type="pull_request" data-hovercard-url="/anthropics/anthropic-sdk-typescript/pull/230

<h2>0.112.3 (2026-07-17)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.112.2...sdk-v0.112.3">sdk-v0.112.2...sdk-v0.112.3</a></p> <h3>Chores</h3> <ul> <li><strong>docs:</strong> small updates (<a href="https://github.com/anthropics/anthropic-sdk-typescript/commit/79fd6c7dd046e066965aa7ab9897695c174f88bb">79fd6c7</a>)</li> </ul>

<h2>0.112.2 (2026-07-17)</h2> <p>Full Changelog: <a href="https://github.com/anthropics/anthropic-sdk-typescript/compare/sdk-v0.112.1...sdk-v0.112.2">sdk-v0.112.1...sdk-v0.112.2</a></p> <h3>Chores</h3> <ul> <li><strong>client:</strong> docs updates (<a href="https://github.com/anthropics/anthropic-sdk-typescript/commit/fdb3a65501a809b4e949131f2097dc0d84e01cc7">fdb3a65</a>)</li> </ul>

Vercel Sandbox no longer bills for data it downloads from the internet. Installing packages, cloning a Git repository, or pulling artifacts and datasets from an external source does not count toward Sandbox Data Transfer usage. Traffic received on a Sandbox's exposed ports is still billable, as is outbound traffic the Sandbox sends to the internet. Pricing for Active CPU, Provisioned Memory, Snapshot Storage, and Sandbox Creations is unchanged. See the pricing documentation for the full breakdow

Runtime logs now show a Cache Reason explaining why a request wasn't a fresh cache hit, for example a time-based or tag-based revalidation. Use cache reasons to debug misses and improve your hit rate. Cache reasons appear for any response the CDN can cache, including ISR, Partial Prerendering, and functions that set a Cache-Control header with directives like stale-while-revalidate . Responses rendered dynamically on every request don't have a cache reason. Cache status Possible reasons MISS Col