One command, 28 regions
vega deploy ships your API to every
region at once. Traffic routes to the nearest healthy edge — 42ms p99 worldwide.
Vega gives your team a production-grade API platform — global deploys, built-in caching, and observability that explains itself. No YAML. No ops rotation.
$ npm install @vega/sdk
import { vega } from '@vega/sdk';
export const api = vega.app({ region: 'global' });
api.get('/users/:id', async ({ params, cache }) => {
const user = await cache.swr(`user:${params.id}`, () =>
db.users.find(params.id),
);
return Response.json(user);
}); $ vega deploy --watch
Compiling api/index.ts…
✓ Built in 1.2s
✓ Deployed to 28 regions
→ https://api.vega.dev (42ms p99)
watching for changes
Trusted by teams shipping 8B requests/month
Platform
Vega owns the run-and-operate layer — deploys, caching, tracing, rollbacks — so your repository stays application code.
vega deploy ships your API to every
region at once. Traffic routes to the nearest healthy edge — 42ms p99 worldwide.
cache.swr() serves stale instantly and
revalidates in the background. Hits return in 6ms at the median, no invalidation service
to run.
Every pull request gets an isolated environment with its own URL. Live in about 10 seconds, deleted on merge.
Every request is traced end to end — middleware, cache, database. No agents to install, no sampling surprises.
Routes, params, and responses are fully inferred in
@vega/sdk. If it compiles, the
contract holds — across every handler and every consumer.
Every deploy is immutable. Pin traffic to any previous version in under 10 seconds globally — no rebuild, no redeploy, no incident channel archaeology.
Developer experience
Caching and auth are function calls, not config files. Declare them next to the route they protect, and Vega runs them at the edge in all 28 regions.
import { vega } from '@vega/sdk';
export const api = vega.app();
api.get('/products', async ({ cache }) => {
const products = await cache.swr('products', fetchProducts, {
ttl: 60,
stale: 300,
});
return Response.json(products);
}); import { vega } from '@vega/sdk';
export const api = vega.app();
api.use(async (ctx, next) => {
const auth = ctx.request.headers.get('authorization');
if (!auth?.startsWith('Bearer ')) {
return new Response('unauthorized', { status: 401 });
}
ctx.trace.set('auth.scheme', 'bearer');
return next();
}); Observability
Vega traces each request through middleware, cache, and database — and streams structured logs alongside. When something is slow, you know which span, in which region, and why.
Tracing is on by default for every Pro deploy — no agent, no sidecar, no sampling config.
Customers
“We shut down our gateway, our cache cluster, and the on-call rotation that babysat them. p99 went from 180ms to 46ms in the first week.”
“The tracing is the feature. A query that had been slow for months showed up on the first waterfall — we fixed it in an afternoon.”
“Preview environments changed how we review. Every pull request ships with a URL, so reviewers test the API instead of reading diffs.”
The Free tier includes 100k requests a month across 3 regions. No credit card, no trial clock — upgrade when your traffic does.