Build your media platform
on Cloudflare's
global network

S3-compatible object storage with zero egress fees. Store raw footage, rendered exports, audio files, project archives, and podcast episodes without worrying about bandwidth bills. Works natively with Workers bindings and integrates with Stream and Images.

Dynamically resize, optimize, crop, and convert images at the edge — no manual variant generation. Serve WebP or AVIF to supported browsers automatically. Bring your own R2 bucket or use fully managed hosted storage.

// Resize + convert to WebP in one URL
/cdn-cgi/image/width=1200,format=webp
/r2/shoots/campaign-hero.jpg

Serverless JavaScript/TypeScript runtime deployed across 330+ cities in milliseconds. Write API logic, authentication handlers, webhook processors, and background jobs — with native bindings to R2, D1, KV, Stream, and Workers AI. No servers, no ops, pay only for what you use.

export default {
  async fetch(req, env) {
    // Verify token, then stream from R2
    const asset = await env.MEDIA_BUCKET
      .get('footage/raw/shoot-42.mp4');
    return new Response(asset.body);
  }
};

Git-connected full-stack deployment on Cloudflare's CDN. Push to GitHub, get a global deployment in seconds. Pages Functions let you run server-side logic alongside your frontend with the same Workers runtime — so your Next.js or SvelteKit app has one unified deployment target.

Run 50+ open-source ML models on serverless GPUs — no GPU provisioning, no model hosting overhead. Generate transcripts, auto-caption videos, classify content, generate image descriptions, or build an AI-powered search index for your media library. Pay per inference, not per hour.

const transcript = await env.AI.run(
  '@cf/openai/whisper',
  { audio: audioBuffer }
);
// Store to D1, surface in your UI
await env.DB.prepare(
  'INSERT INTO transcripts VALUES (?,?)'
).bind(videoId, transcript.text).run();

Each Durable Object is a globally-unique, named unit of compute with strongly consistent storage co-located alongside it. Unlike regular Workers, they maintain in-memory state between requests and coordinate across multiple clients — making them the backbone of AI agents, real-time collaboration, live notifications, and review workflows that span sessions.

export class ReviewAgent {
  constructor(state, env) {
    this.state = state; // persistent SQLite
  }
  async fetch(req) {
    const { action, note } = await req.json();
    await this.state.storage.put('status', action);
    // Notify all connected clients via WS
    this.broadcast({ action, note });
  }
}

Run headless Chrome on Cloudflare's global network — no browser infrastructure to manage. Capture screenshots, generate PDFs, scrape structured data, and run full Puppeteer or Playwright scripts from inside a Worker. Pair with Durable Objects to persist sessions across requests and cut cold-start time. Supports AI-powered browsing via Playwright MCP and Stagehand.

// One HTTP request — no browser to manage
const pdf = await fetch(
  'https://api.cloudflare.com/client/v4/accounts'
  + '/{id}/browser-run/pdf',
  {
    method: 'POST',
    body: JSON.stringify({ url: proofPageUrl }),
    headers: { 'Authorization': 'Bearer TOKEN' }
  }
);