Platform

Everything your AI app needs.
In one platform.

Deploy your app, serve its model and grow both from one place.

Deploy

Push code. Get a live link.

Connect a repo. Kordin builds it and puts it online.

GitHub repositories

Connect a repo through the GitHub app.

Builds

Buildpacks by default. Your Dockerfile when you need it.

CLI and MCP

So coding agents can deploy for you.

Deploy-to-Kordin button

Templates come with a “Deploy this template to Kordin” button.

Config and secrets

Settings per project and service. Secrets are write-only.

Your own database

Connect Neon, Supabase, RDS and others, with one-click templates for Neon and Supabase. There are no managed databases yet.

Models

Pick a model. Get an API.

We size it and serve it on your GPU.

Model catalogue

Start from models we’ve prepared.

Hugging Face models

Bring any supported Hugging Face model, pinned to a revision, subject to GPU availability.

Sizing check

We check the fit and size the GPU before you deploy.

One-click deploy

Put a model on your reserved GPU in one click.

API endpoint

Each model answers on its own HTTPS endpoint.

Rollback

Roll a model back, just like an app.

GPUs

Whole GPUs. All yours.

No other customer runs on your GPU. We run the serving.

Whole GPUs

No other customer runs on your GPU.

Several models, one GPU

Run several compatible models on the same GPU, if memory is available.

Sets of 1, 2, 4 or 8

GPUs come in sets of 1, 2, 4 or 8.

On demand or reserved

Rent by the hour or reserve for a term.

Usage records

See usage, including idle time.

Serving run for you

We run the serving stack on your reserved GPUs.

Network

Links that work. From the first deploy.

Every service gets a hostname with TLS.

HTTPS hostname

Every service gets a hostname with TLS.

Custom domains

Bring your own domain on any paid plan, from $9 on Founding.

Private routes

Your app and model talk over private project routes.

Load balancing

Traffic spread across ready instances.

Free bandwidth

Ingress and egress are free on every paid plan, fair use. The Free plan includes 50 GB a month.

Dedicated public IPv4

$10 a month, on any paid plan.

Scale preview

Kubernetes. When you need it.

Grow without moving platforms.

support-bot · productionILLUSTRATION · EXAMPLE COMMANDS
$ kordin kube credentials --env production✓ Scoped credentials issued · short-lived$ kubectl get nodesNAME             POOL      STATUSpool-general-1   general   Readypool-general-2   general   Readypool-gpu-1       h200      Ready$ kubectl get pods -n support-botNAME                 READY   STATUSsupport-bot-6c9f     1/1     Runningqwen-3-8-27b-0       1/1     Running$ 
generalmin 2 · max 6
h200min 1 · max 4

WORKER POOLS · ILLUSTRATION

Kubernetes runtime

One per environment, managed and upgraded for you.

Worker pools

Set each pool’s minimum and maximum.

Scoped kubectl

Short-lived credentials from the CLI.

Audit export

Export your audit trail.

Operate

See what happened. Undo it.

Search your logs, follow each step and go back.

Build and deploy logs

Searchable, with a live tail.

Operation events

Progress, and a clear reason when something fails.

Rollback

Go back to an earlier build.

Safe deletion

Delete an environment and its runtime cleanly.

Secure

Built to pass your review.

Your builds, GPUs and secrets stay yours.

SOC 2 Type 1

Independent audit report, on request.

Cyber Essentials

UK government-backed certification.

Isolated builds

Every build runs on its own.

Tenant-only GPUs

Your GPU is never scheduled for anyone else.

Write-only secrets

Stored versioned. Never shown again.

GDPR data processing agreement

Free on every plan.

Coming next

On the roadmap.

Preview environmentsWorkers and cronManaged databasesPersistent volumesMetrics and alertsSAML SSOSCIMBring your own cloudExport and migration

Features describe the first release as planned. We confirm availability as early access opens.

Ship your
first app.

Early access by invitation.

We’ll email your invite as places open.

We’ll email your invite as places open.