The Cloudflare Blog

We built a custom technology stack to run fast large language models on Cloudflare’s infrastructure. This post explores the engineering trade-offs and technical optimizations required to make high-performance AI inference accessible.

We're building AI Gateway into a unified inference layer for AI, letting developers call models from 14+ providers. New features include Workers AI binding integration and an expanded catalog with multimodal models.

Give your agents, developers, and automations a home for code and data. We’ve just launched Artifacts: Git-compatible versioned storage built for agents. Create tens of millions of repos, fork from any remote, and hand off a URL to any Git client.

Learn how to deploy PlanetScale Postgres and MySQL databases via Cloudflare and connect Cloudflare Workers.

AI Search is the search primitive for your agents. Create instances dynamically, upload files, and search across instances with hybrid retrieval and relevance boosting. Just create a search instance, upload, and search.

Agents are becoming multi-channel. That means making them available wherever your users already are — including the inbox. Today, Cloudflare Email Service enters public beta with the infrastructure layer to make that easy: send, receive, and process email natively from your agents.

Announcing a preview of the next edition of the Agents SDK — from lightweight primitives to a batteries-included platform for AI agents that think, act, and persist.

An experimental voice pipeline for the Agents SDK enables real-time voice interactions over WebSockets. Developers can now build agents with continuous STT and TTS in just ~30 lines of server-side code.

Cloudflare Workflows, a durable execution engine for multi-step applications, now supports higher concurrency and creation rate limits through a rearchitectured control plane, helping scale to meet the use cases for durable background agents.

Browser Rendering is now Browser Run, with Live View, Human in the Loop, CDP access, session recordings, and 4x higher concurrency limits for AI agents.

The Cloudflare Registrar API is now in beta. Developers and AI agents can search, check availability, and register domains at cost directly from their editor, their terminal, or their agent — without leaving their workflow.

Agent Lee is an in-dashboard agent that shifts Cloudflare’s interface from manual tab-switching to a single prompt. Using sandboxed TypeScript, it helps you troubleshoot and manage your stack as a grounded technical collaborator.

Managed OAuth for Cloudflare Access helps AI agents securely navigate internal applications. By adopting RFC 9728, agents can authenticate on behalf of users without using insecure service accounts.

We share Cloudflare's internal strategy for governing MCP using Access, AI Gateway, and MCP server portals. We also launch Code Mode to slash token costs and recommend new rules for detecting Shadow MCP in Cloudflare Gateway.

Cloudflare is introducing scannable API tokens, enhanced OAuth visibility, and GA for resource-scoped permissions. These tools help developers implement a true least-privilege architecture while protecting against credential leakage.

Cloudflare Mesh provides secure, private network access for users, nodes, and autonomous AI agents. By integrating with Workers VPC, developers can now grant agents scoped access to private databases and APIs without manual tunnels.

We’re introducing cf, a new unified CLI designed for consistency across the Cloudflare platform, alongside Local Explorer for debugging local data. These tools simplify how developers and AI agents interact with our nearly 3,000 API operations.

Cloudflare Sandboxes give AI agents a persistent, isolated environment: a real computer with a shell, a filesystem, and background processes that starts on demand and picks up exactly where it left off.

We’re introducing Durable Object Facets, allowing Dynamic Workers to instantiate Durable Objects with their own isolated SQLite databases. This enables developers to build platforms that run persistent, stateful code generated on-the-fly.

Outbound Workers for Sandboxes provide a programmable, zero-trust egress proxy for AI agents. This allows developers to inject credentials and enforce dynamic security policies without exposing sensitive tokens to untrusted code.