Introducing Forge v2: Edge-First APIs

Today we're announcing Forge v2 - a complete rebuild of our API platform from the ground up. Every request now routes through our global edge network, delivering 23ms median response times across 18 regions worldwide.

Why we rebuilt everything

When we launched Forge in 2022, we ran on a single-region architecture. It was simple, fast to iterate on, and served us well to our first 800 customers. But as our customers started deploying globally, the latency from a single US-East region became untenable.

We evaluated several approaches: multi-region replication, edge caching layers, and CDN-based routing. Ultimately, we chose to push our entire compute layer to the edge, running on a custom runtime that boots in under 5ms.

const apiOrigin = "https://api.replace-with-your-domain.example.com";

// Before: Single region
const response = await fetch(new URL("/v1/events", apiOrigin).toString());
// Latency: 180ms from Tokyo, 95ms from London

// After: Edge-first
const response = await fetch(new URL("/v2/events", apiOrigin).toString());
// Latency: 12ms from Tokyo, 8ms from London

What's new in v2

Beyond the infrastructure overhaul, v2 introduces several features our customers have been requesting: streaming responses for large payloads, native WebSocket support for real-time data, cursor-based pagination across all list endpoints, and a new batch API for high-throughput workloads.

The new TypeScript SDK (v3) takes full advantage of these improvements. Response types are inferred from the endpoint path — no codegen, no type assertions, just pure type safety from request to response.

Migration guide

We've designed v2 to be backwards compatible with v1 for all core endpoints. The /v1 prefix will continue to work through 2026. To start using v2 features, update your SDK to version 3.x and change your base URL to /v2. Full migration guide is available in our documentation.