No servers, no nodes. Just your container.
Run your Docker image in Istanbul. Pick 0.25-4 vCPU and 0.5-30 GB of memory per container and pay only while it runs. HTTPS address, autoscaling and alerts included.
- Runs in Istanbul (tr-ist-1)
- Metered by the minute
- Every container in its own isolation layer
- API · CLI · Terraform
Cost calculator25% off
Excl. VAT. Charged hourly from your balance; nothing is charged while copies are at 0. €1 = 56.0719 TRY (today's rate).
Start with this size- 0.25 vCPUsmallest container; up to 4 vCPU / 30 GB
- Per minutemetering, hourly charging, no monthly cap
- ~15 sfrom image to HTTPS address (measured, capacity ready)
- €0charged while a service is scaled to 0
Serverless or dedicated nodes?
Both modes share the same panel, HTTPS address, API, CLI and Terraform. The difference is what you pay for.
Serverless
Pay per container
- Small services, APIs, test and preview environments
- Spiky traffic, or traffic at certain hours of the day
- vCPU / memory per container, no server plan
- vCPU-hours + GB-hours; no monthly cap
Dedicated nodes (ECS)
Nodes reserved for you
- Steady, predictable load and many services at once
- Nodes are yours alone; containers fill them as they fit
- Billed hourly, capped at the plan price per month
- Smallest plan: at most 391.94 TL (€6.99) per month
Rule of thumb: if the same resources run around the clock all month, a node plan is usually cheaper; for short-lived or spiky work, serverless saves money.
Need root access or a persistent disk? Gyro Cloud Istanbul virtual servers
Live in four steps
No servers to set up, no cluster to manage. Give us your image and a size, the platform does the rest.
- Step 1
Top up and order
Prepaid. About one day of the smallest container in your balance is enough to order; there is no environment fee.
- Step 2
Pick a size
From 0.25 vCPU / 0.5 GB up to 4 vCPU / 30 GB. The hourly estimate is shown under the form.
- Step 3
Publish the image
From Docker Hub, GitHub or your private Gyro registry. In seconds you get a service-code.gyroapp.net address with a certificate.
- Step 4
It scales, it is metered
Copies are added as load grows and removed as it drops. Each running copy is metered by the minute and charged hourly.
Pay as you go, per copy
Hourly price = vCPU × vCPU-hour + GB × GB-hour. The amounts below assume one copy running 24/7; containers that run less pay less.
- vCPU-hour€0.0298
€0.0398 - GB-hour€0.0024
€0.0032 - 25% off
| vCPU | Memory | Per hour | Full month (730 h) |
|---|---|---|---|
| 0.25 | 0.5 GB | 0.4850 TL (€0.0087) | |
| 0.25 | 1 GB | 0.5523 TL (€0.0099) | |
| 0.5 | 1 GB | 0.9700 TL (€0.0173) | |
| 1 | 2 GB | 1.94 TL (€0.0346) | |
| 2 | 4 GB | 3.88 TL (€0.0692) | |
| 2 | 8 GB | 4.42 TL (€0.0788) | |
| 4 | 8 GB | 7.76 TL (€0.1384) | |
| 4 | 30 GB | 10.72 TL (€0.1912) |
Excl. VAT, TRY amounts at today's rate. Memory options per vCPU: 0.25 → 0.5-2 GB · 0.5 → 1-4 GB · 1 → 2-8 GB · 2 → 4-16 GB · 4 → 8-30 GB (1 GB steps). Image registry: 2 GB free, then €0.05 per GB-month.
Up to 16 vCPU / 32 GB running at once per account (default); open a support ticket for more and we raise it for your account.
Serverless, yet complete
Everything in the node-based product is available in serverless mode too.
Automatic HTTPS address
Each service gets a service-code.gyroapp.net address and certificate automatically.
Your own domain + SSL
Point your domain with a CNAME; the certificate is issued and renewed automatically.
Autoscaling
Set min / max copies and a CPU target; copies come and go with the load.
Alerts
E-mail, SMS, Telegram or webhook when a service goes down, keeps restarting or hits its maximum.
Zero-downtime releases, one-click rollback
New versions roll out copy by copy; the last 15 revisions are kept and you can roll back with one command.
Logs and metrics
Live logs, the last output of a crashed copy, and the service's CPU and memory use.
Private image registry
A private registry for your environment with 2 GB free. Credentials for external registries can be set too.
Variables and secrets
Environment variables and secrets are managed separately; changes roll out as a new revision without downtime.
Health checks
HTTP or TCP health checks; copies that are not ready get no traffic, crashed copies restart.
Every container in its own isolation layer
Serverless containers run on shared capacity but cannot see each other. Each copy runs in its own kernel sandbox (open-source gVisor) that handles system calls before they reach the host.
- Non-root user; extra capabilities dropped, no privileged mode.
- No access to other customers' containers or to infrastructure networks; outbound SMTP 25 closed (587 open).
- CPU, memory and disk are capped at the size you choose; a noisy neighbour cannot slow you down.
- In our own test, a neighbour service answered 25 of 25 requests while the container next to it was under a process flood.
Grows with traffic, shrinks after it, and so does the bill
Set the minimum and maximum copies and a CPU target per service. In serverless mode only running copies are charged, so a quieter day costs less.
- When average use goes over the target, copies can double within 30 seconds.
- When load drops, after 3 quiet minutes one copy is removed per minute.
- Copies that hit the account limit or regional capacity wait for room and you get a "capacity" alert.
- No traffic at night? Set the minimum to 1, or scale an unused service to 0: zero copies cost nothing.
Be the first to know when something goes wrong
Five event types, four channels. An event is announced when it lasts at least two scans (about 2 minutes), it is not repeated within the cooldown, and a "Resolved" message follows when it is over.
Events
- Service down (no ready replica)
- Keeps restarting / image cannot be pulled
- Ready replicas below your threshold
- Autoscaling at its maximum
- Account limit or regional capacity reached
Channels
- SMS
- Telegram
- Signed webhook
Gyro Kapsayıcı · #17521
web: autoscaling reached its maximum
The service is at its maximum of 3 copies and load is still above target. Consider raising the maximum or the container size.
Gyro Kapsayıcı · #17521
Resolved: web: autoscaling reached its maximum
The condition is back to normal.
Dear Jane Doe,
State: CrashLoopBackOff, restarts: 5. See Logs > previous instance for the crashed replica output.
Open serviceWebhook body (signed JSON)
POST https://ornekfirma.com/hooks/gyro
X-Gyro-Signature: sha256=9f2c…
{
"event": "container.alert.trigger",
"service": "api", "type": "down", "status": "firing",
"title": "Service api is down",
"message": "No ready replica (0/2). Check the latest error on the Logs page."
}
Size is two parameters: --cpu-m and --memory-mi
Serverless mode uses the same API endpoint, CLI commands and Terraform resource as the node-based product; you just set the size.
# curl -fsSL https://www.gyrohosting.com/cli/install.sh | sh
gyro configure
# 0.25 vCPU / 0.5 GB, HTTPS address
gyro container service create 12345 --name api \
--image ghcr.io/acme/api:1.0 --port 8080 \
--cpu-m 250 --memory-mi 512 --health-path /healthz --public
# Bigger size: 1 vCPU / 2 GB (zero-downtime rollout)
gyro container service update 12345 api --cpu-m 1000 --memory-mi 2048
# Autoscaling 1-5 copies at 70% CPU
gyro container service autoscale 12345 api \
--autoscaling '{"enabled":true,"min":1,"max":5,"cpu_percent":70}'
# Not in use? 0 copies = no charge
gyro container service scale 12345 api --replicas 0B=https://www.gyrohosting.com/api/v1/client
H=(-H "Authorization: Bearer $GYRO_TOKEN" -H "Content-Type: application/json")
# Create a service: 0.25 vCPU / 0.5 GB
curl -s -X POST "$B/cloud/containers/12345/services" "${H[@]}" -d '{
"name": "api", "image": "ghcr.io/acme/api:1.0", "port": 8080,
"cpu_m": 250, "memory_mi": 512, "health_path": "/healthz", "public": true}'
# Resize: 1 vCPU / 2 GB
curl -s -X PATCH "$B/cloud/containers/12345/services/api" "${H[@]}" \
-d '{"cpu_m": 1000, "memory_mi": 2048}'
# Scale to 0 (no charge)
curl -s -X POST "$B/cloud/containers/12345/services/api/scale" "${H[@]}" \
-d '{"replicas": 0}'terraform {
required_providers {
gyro = { source = "gyro-hosting/gyro" }
}
}
resource "gyro_container_service" "api" {
service_id = 12345 # serverless service no.
name = "api"
image = "ghcr.io/acme/api:1.0"
port = 8080
cpu_m = 250 # 0.25 vCPU
memory_mi = 512 # 0.5 GB
public = true
health_path = "/healthz"
autoscaling = { min = 1, max = 5, cpu_percent = 70 }
}
output "url" { value = gyro_container_service.api.url }All endpoints, authentication and samples are in the developer docs. Developer docs
Step-by-step guides
- Serverless mode: sizes, pricing and limits
- Getting started: publish your first app in 5 minutes
- First app: image, port, health check and replicas
- Gyro image registry: docker login, push, clean-up
- Your own domain and free SSL
- Autoscaling and alerts
- Manage with the API, gyro CLI and Terraform
- Automatic releases with GitHub Actions
- Limits and rules: SMTP, resources, disk
What are serverless containers?
How is it billed?
Where do my containers run?
Which sizes can I choose?
How is it different from node-based Gyro Containers?
How many containers can I run at once?
Why did my first start take a while?
Is there a persistent disk?
Can I run images that need root?
Can I use my own domain?
What happens if my balance runs out?
Can my container send e-mail?
Your first container is minutes away
Top up, give us your image, pick a size. No charge for hours you do not use.
Not sure which plan fits you?
Tell us what you need and our team will recommend the right setup within minutes.
Reach Sales & Support
Have a question before you buy, or need help with an existing account? Call or email us.