Edge platform for real-time apps

The backend that lives
next to your users.

Functions, a database, queues and storage on one network of 38 regions. Deploy once and every request is answered close to where it started.

Start building free Read the docs

The Lattice console: every request, region and deploy in one place.

Built for teams like

One platform from first request to last byte.

Every part of your backend gets the same global footprint, so you stop stitching vendors together and get back to shipping.

Bring your framework
Deploy Node, Bun, Python or Go handlers with the tools and libraries you already use.
Streams by default
Responses, queries and events stream as they happen, with no extra setup.
Global from day one
Code runs in the region closest to each request, automatically.
Pay per request
Usage pricing with no idle servers and no long-term contract to sign.
Dedicated capacity
Reserve isolated compute for steady, high-volume workloads.
Private networking
Reach your own cloud over private links in every region we run.

Explore the platform

Building blocks

Six primitives. One bill.

Start with one, add the rest when you need them. They share a network, a dashboard and an invoice.

See all products

Edge Functions

Code runs in 38 regions and starts in under 10 ms.

View docs →

Global Database

Postgres-compatible reads from the nearest replica, writes kept consistent.

View docs →

Queues

Durable jobs and retries that survive deploys and region outages.

View docs →

Object Storage

S3-compatible buckets with no fees for data going out.

View docs →

Realtime

Push updates to millions of open sockets from one API call.

View docs →

Inference

Call open models from your functions, in region, by the token.

View docs →

Closer is faster, by a lot.

Median response times from twelve client cities, Lattice against the same app on a single cloud region.

Source: Lattice lab test, October 2026. Demo figures for this template.

API read

Lattice, 38 regions9 ms

Single cloud region84 ms

Database write

Lattice, 38 regions14 ms

Single cloud region120 ms

Cold start

Lattice, 38 regions6 ms

Single cloud region410 ms

Realtime fan-out, 10k clients

Lattice, 38 regions38 ms

Single cloud region260 ms

Pick a workload. See what it runs on.

The same blocks become very different products. Choose one to see the pieces it uses and the code that ties them together.

A chat backend in 12 lines: store the message, send it to the room, done.

Chat and messaging

Runs on Functions, Realtime, Database and Queues. Messages land in under 40 ms for both people, wherever they are.

Open the chat template →

A live dashboard: watch a query and pipe it to a channel.

Live dashboards

Runs on Functions, Database, Realtime and Storage. Rows stream from Postgres to every open screen as they change.

Open the dashboard template →

Presence and shared edits for any canvas or document.

Multiplayer apps

Runs on Functions, Realtime, Database and Storage. Cursors and edits reach everyone in the room at once.

Open the multiplayer template →

A checkout handler: write the order, queue the receipt.

Checkout and carts

Runs on Functions, Database, Queues and Storage. Orders are written once and receipts retry until they land.

Open the checkout template →

An ingest endpoint that answers in milliseconds and queues the rest.

Device telemetry

Runs on Functions, Queues, Database and Realtime. Thousands of sensors report into one live view.

Open the telemetry template →

Start from something that already works.

Clone a template, push to main and it is live in every region. Change what you like after that.

ComputeChat backend Rooms, history and push notifications in one deploy. DataLive dashboard Stream rows from Postgres straight to the browser. RealtimeMultiplayer cursors Presence and shared state for any canvas or doc. AISupport agent Answer tickets with an in-region model and your docs.

Our network, end to end.

38Regions on hardware we own

4.2BRequests answered every day

212Peering partners on our backbone

99.99%Uptime over the last 12 months

Three layers, and we own all of them.

A competitor can match a price. Matching the stack under it takes years.

How the platform fits together →

01layer

Compute

Functions, actors and scheduled jobs on every region’s own servers.

Explore compute →

02layer

Data

Database, queues and storage, kept right next to your code.

Explore data →

03layer

Network

Our own backbone, anycast routing and peering in 38 cities.

Explore network →

Regions

Find the region next door.

Every region runs the full platform. Requests go to the closest one and fail over on their own.

11 regions, from Ashburn to Vancouver.
9 regions, including two in the EU sovereign zone.
12 regions, Tokyo to Sydney.
6 regions, São Paulo to Johannesburg.

What you get, side by side.

Most teams run a cloud region and five services on top. Here is the same app on each setup.

Time to first deploy

From an empty folder to live traffic.

Lattice: 2 min

Single cloud region: 1 day

Build it yourself: 2 weeks

Program comparison
Compare	Lattice	Single cloud region	Build it yourself
Regions includedWhere your code runs on day one.	Lattice✓ 38	Single cloud region1	Build it yourselfYour call
Database replicasRead copies close to every user.	Lattice✓ Built in	Single cloud regionExtra service	Build it yourselfYou run it
Realtime socketsOpen connections to browsers and apps.	Lattice✓ Built in	Single cloud regionThird party	Build it yourselfYou run it
Data going outWhat you pay when users download.	Lattice✓ Free	Single cloud regionPer GB	Build it yourselfPer GB
Private networkingLinks into your own cloud.	Lattice✓ Every region	Single cloud regionSome regions	Build it yourselfYou build it

Price your workload before you write a line.

Set your expected usage and see the monthly bill. One rate card covers every product.

RequestsMillions of function and API calls a month

Database storageGigabytes kept across all replicas

200

Realtime connectionsThousands of open sockets at peak

Inference tokensMillions of model tokens a month

For agents

Built for agents and the people who write them.

Every product on this page is also an API. Point your coding agent at Lattice and it can deploy, query and roll back on its own.

CLI

$ lattice init chat-app
$ lattice deploy
  ✓ built in 3.1 s
  ✓ live in 38 regions
  → https://chat-app.lattice.run

SDK

import { db, live } from "lattice"

const room = live.room("419")
room.on("message", async (m) => {
  await db.insert("messages", m)
})

MCP server

{ "mcpServers": {
    "lattice": {
      "url": "https://mcp.lattice.dev"
} } }

// tools: deploy, logs, query, rollback

Bring your framework

Streams by default

Global from day one

Pay per request

Dedicated capacity

Private networking

Six primitives. One bill.

Edge Functions

Global Database

Queues

Object Storage

Realtime

Inference

Closer is faster, by a lot.

Chat and messaging

Live dashboards

Multiplayer apps

Checkout and carts

Device telemetry

Start from something that already works.

Our network, end to end.

Three layers, and we own all of them.

Compute

Data

Network

Regions

Find the region next door.

Price your workload before you write a line.

For agents

Built for agents and the people who write them.

CLI

SDK

MCP server

Deploy your first functionin two minutes.

Resources

Deploy your first function
in two minutes.