Skip to content
WEBIS Runtime

v4.2 — 31 edge regions live

Ship edge functions without the ops team.

Push to Git and your functions are live in 31 regions in under 900 ms. No cluster to size, no YAML to memorise, and logs you can actually read.

No credit card Free tier: 1M requests / mo SOC 2 Type II

Trusted by engineering teams at

  • Northpeak
  • Corelayer
  • Blockwise
  • Vertexa
  • Halden
  • Quartzly

01 — Platform

Everything the runtime handles so you don't have to

Routing, scaling, secrets, and observability ship in the box. You write the handler.

Instant global routing

Requests land in the region closest to the user. Failover is automatic and takes 40 ms.

  • icn1 · Seoul12 ms
  • nrt1 · Tokyo24 ms
  • sin1 · Singapore58 ms
  • fra1 · Frankfurt141 ms
  • iad1 · Virginia168 ms

Sample latencies from a Seoul client · illustrative

Secrets that stay secret

Encrypted at rest, injected at cold start, never printed in logs.

Logs you can read

Structured by default, searchable for 30 days, no query language to learn.

Zero-config scaling

One request or a million — the same handler, no autoscaler to tune.

Preview per branch

Every pull request gets its own URL with production-shaped data.

One-click rollback

Every deploy is immutable. Roll back to any of the last 100 in 2 seconds.

02 — Speed

Cold starts measured in single digits

The runtime keeps a warm pool per region and snapshots your isolate after the first request. Most handlers never see a cold start after deploy minute one.

Median cold start
7ms
Deploy to live
880ms
Edge regions
31
Uptime, 12 mo
99.99%
handler.ts
// no config file, no Dockerfile
export default {
  async fetch(req: Request) {
    const city = req.headers.get("x-webis-city");
    return Response.json({
      hello: city ?? "world",
      region: WEBIS.region,
    });
  },
};
$ webis deploy live in 880ms →

03 — Workflow

Three steps, then it's just Git

  1. STEP 01

    Connect the repo

    Point us at a GitHub or GitLab repository. We detect the runtime and entry file automatically.

  2. STEP 02

    Push a branch

    Every branch gets a preview URL. Merge to main and it promotes to production automatically.

  3. STEP 03

    Watch it run

    Latency, errors, and logs stream into one dashboard. Alerts go to Slack or a webhook.

04 — Use cases

What teams put on the edge first

Personalisation at the CDN layer

Rewrite HTML per visitor — locale, currency, A/B bucket — before it leaves the edge. Your origin keeps serving one cached document, so hit rates stay above 95%.

  • → Geo and locale headers included on every request
  • → Deterministic bucketing helper built in
  • → Streams HTML, so TTFB stays flat

API gateway and auth

Verify tokens, enforce rate limits, and fan out to internal services from a single handler. Rejected traffic never reaches your origin or your bill.

  • → JWT and JWKS verification in the standard library
  • → Per-key rate limiting with regional counters
  • → Request coalescing for hot upstreams

Webhook intake that never drops

Accept, validate, and queue provider callbacks in the nearest region. Bursts of 40k webhooks a minute land without a warm-up window.

  • → Signature verification helpers for common providers
  • → Durable queue handoff with at-least-once delivery
  • → Replay any message from the last 7 days

Image and asset transforms

Resize, re-encode, and watermark on first request, then cache the result at the edge forever. Your bucket stores one original per asset.

  • → AVIF and WebP negotiation from Accept headers
  • → Signed transform URLs to stop hotlink abuse
  • → Immutable cache keys per transform recipe

05 — Comparison

What you stop maintaining

A typical container setup versus the same workload on the runtime.

Container platform compared with WEBIS Runtime
Concern Container platform WEBIS Runtime
Capacity planning Pick instance sizes, set replica counts, tune the autoscaler None — isolates spin per request
Global presence Duplicate the stack per region, manage failover yourself 31 regions on by default
Cold start Seconds — image pull, boot, health check 7 ms median from snapshot
Idle cost Minimum replicas bill around the clock Zero — you pay per request
Rollback Re-deploy a previous tag, wait for the rollout Two seconds, any of 100 revisions
Config surface Dockerfile, manifests, ingress, HPA, secrets store One handler file

Illustrative comparison for a sample workload. Your numbers will differ.

06 — Ecosystem

Connects to what you already run

Browse all integrations →
  • Postgres
  • Redis
  • Object storage
  • Kafka
  • GitHub
  • GitLab
  • Slack alerts
  • OpenTelemetry
  • Terraform
  • Secrets manager
  • Sentry
  • Payments

07 — Security

Isolation is the default, not a tier

Every request runs in its own V8 isolate with no shared filesystem and no ambient network access. Outbound calls go through an allowlist you control.

SOC 2 Type II ISO 27001 GDPR K-ISMS ready
Per-request isolation
No process reuse between tenants. A crash affects exactly one request.
Egress allowlist
Declare the hosts a function may reach. Anything else is refused at the socket.
Secret redaction in logs
Injected values are masked before a log line is written, including inside stack traces.
Immutable audit trail
Deploys, rollbacks, secret reads, and member changes are recorded and exportable.
Region pinning
Constrain execution and log storage to a single jurisdiction when policy requires it.

08 — Teams

What changed after the move

"We deleted 4,000 lines of Terraform the week we moved. Nobody has missed it."
Sena ParkPlatform Lead, Northpeak
"Preview URLs per pull request killed our staging queue. Reviews got faster overnight."
Daniel KwonStaff Engineer, Corelayer
"Our p95 in Seoul went from 340 ms to 61 ms and the bill went down. That never happens."
Mina YooCTO, Blockwise

09 — Pricing

Pay per request, not per cluster

No seat minimums. No idle charges. Stop deploying and the bill stops.

Hobby

Side projects and prototypes

$0

forever

  • 1M requests / month
  • 3 edge regions
  • 7-day log retention
Start free
Most picked

Team

Production workloads

$40/mo

+ $0.30 per extra million

  • 25M requests included
  • All 31 regions
  • 30-day logs + alerts
  • Preview per branch
Start 14-day trial

Enterprise

Compliance and scale

Custom

annual agreement

  • Dedicated isolate pools
  • SSO, audit log, SCIM
  • 99.99% SLA, named SRE
Talk to us

10 — FAQ

Questions engineers actually ask

Which runtimes do you support?

TypeScript and JavaScript on the V8 isolate runtime, plus WASM modules compiled from Rust or Go. Long-running processes belong on a container platform, not here.

How is billing calculated?

Per request plus CPU milliseconds actually consumed. Idle isolates cost nothing, so a quiet week costs a quiet week.

Can we keep data inside one country?

Yes. Region pinning is available on Team and Enterprise. Pin to icn1 and requests never leave Korea.

What happens if a deploy breaks production?

Every deploy is immutable and kept for 100 revisions. Roll back from the dashboard or the CLI in about two seconds — no rebuild required.

Is there a migration path off the platform?

Handlers use the standard Request and Response objects, so the same code runs on any compliant runtime. Export your config as JSON whenever you like.

Your first function is live in 90 seconds

Free tier, no card, no sales call. Delete the project any time and nothing lingers.

Demo form — submissions are not sent anywhere.