Scalix Run

Docker containers.
Managed and scaled.

Deploy long-running Docker containers with concurrency-based auto-scaling, revisions, rollbacks, health checks, and scale-to-zero. Pay per vCPU-hour and GiB-hour — only for what you use.

Deploy a service Read the docs

scalix-run — deploy

connected·eu-central

from $0.036/vCPU-hour·scale to zero·concurrency autoscaling·revisions & rollback·SSL included

Run anything

One platform, every workload.

If it packages into a container, Run deploys it. Same URL, SSL, autoscaling, and scale-to-zero for every kind of service.

Web apps & APIs

Serve a frontend or an HTTP API on a stable URL with SSL, scaling up and back down as traffic moves.

Background workers

Run queue consumers, schedulers, and long-lived jobs as their own service, sized to the work in front of them.

AI agents

Host agent loops that call tools, hold state, and keep running for as long as a task takes.

MCP servers

Deploy a Model Context Protocol server as a long-running service that AI clients connect to over a persistent URL.

Bring any Docker image

Push a prebuilt image, bring a Dockerfile, or push source. Node, Python, Go, and Rust build automatically.

Host MCP servers

An MCP server is a long-running service that AI clients connect to over a persistent link, which is exactly the shape Run is built for. Deploy yours with one command and it gets a stable URL, SSL, and scale-to-zero like every other service. New to this? Read what an MCP server is and how to host one.

Elastic by default

Scales with traffic. Sleeps when idle.

Instances are added when request concurrency climbs and removed as it falls — no schedules, no capacity guesswork. When traffic stops entirely, the service scales to zero and the meter stops with it.

Ship without fear

Every deploy is a revision.

Each deploy creates an immutable revision, and health checks gate the cutover — a bad build never takes traffic. If anything slips through, roll back to any previous revision in seconds.

v40

v41

v42 · live

rollback --to v41 · seconds · zero downtime

Pricing

$0.036 per vCPU-hour + $0.004/GiB-hour. Scale to zero = pay nothing when idle.

SizevCPUMemoryHourlyMonthly (24/7)

Nano½ vCPU256 MB$0.019$13.87 (Fly.io shared: $4.10)

Micro½ vCPU512 MB$0.020$14.60 (Fly.io 2x: $8.21)

Standard 1x1 vCPU2 GB$0.044$32.12 (Fly.io perf: $23.33)

Standard 2x2 vCPU4 GB$0.088$64.24 (Cloud Run: ~$31)

Standard 4x4 vCPU8 GB$0.176$128.48 (Cloud Run: ~$62)

One flat rate for every size — $0.036/vCPU-hour and $0.004/GiB-hour, billed per second while a service is awake. Sizes above are the monthly cost of running 24/7; scale to zero and you pay nothing when idle. Competitor figures are public list prices as of June 2026.

Everything a service needs

Any Docker image

Bring a Dockerfile or a prebuilt image. Pushed to the Scalix Registry, deployed with one command.

Health checks

HTTP health checks gate every rollout — a revision only takes traffic once it answers healthy.

Domains & SSL

A run.scalix.world URL out of the box; bring your own domain and certificates are provisioned and renewed automatically.

Log streaming

Live logs per service and per instance, in the console or from the CLI.

Sized to the job

Pick vCPU and memory per service, from shared 256 MB up to 4 vCPU / 8 GB. Change anytime.

Environment variables

Configuration injected per service and per revision — change config without rebuilding the image.

Concurrency-based autoscalingRevision trackingOne-click rollbackScale to zero

Docker containers.Managed and scaled.