Home Features Pricing Docs Contact Demo
Open Source

Find wasted Kubernetes
spend and get
copy-paste fixes.

KubeOpt analyzes your clusters, shows you where money is going, and gives you specific recommendations with the YAML to act on them. Works on AKS, EKS, and GKE.

Or run locally in one command:

$ npx kubeopt demo
localhost:5001
Total monthly cost
$6,960
Savings available
$2,556
Clusters
3
ClusterCostSavingsScore
prod-eks-us-east-1$4,800$1,68061%
staging-gke-eu-west1$1,200$39672%
dev-aks-westus2$960$48048%
Top finding
6 deployments in prod-eks request 3x more CPU than their 30-day peak.
kubectl set resources deployment/api-server --requests=cpu=500m,memory=512Mi

How it works

Connect read-only credentials. Run the analysis. Review recommendations and YAML in the dashboard.

1

Connect

$ npx kubeopt start

Add your cluster and read-only cloud credentials in the dashboard. No agents installed in your cluster.

2

Analyze

$ kubeopt analyze prod-aks

KubeOpt scans pods, nodes, storage, and cloud billing. Findings appear in the dashboard with cost impact and confidence scores.

3

Fix

$ kubeopt report prod-aks

Every recommendation comes with the specific kubectl command or YAML change to apply it. Copy, review, run.

What it finds

Based on real kubectl metrics and cloud billing APIs. Not estimates from averages.

Oversized Pods

Pods requesting 2-4x more CPU and memory than they actually use. KubeOpt calculates right-sized limits and generates the YAML patch.

Idle Workloads

Deployments consuming resources but receiving no traffic. Staging environments left running. Dev namespaces nobody has touched in weeks.

Node Waste

Nodes running at low utilization because bin-packing is off. Wrong VM sizes for your actual workload profile.

Missing Autoscaling

Deployments without HPA that should have it. Fixed replica counts paying for peak-traffic capacity 24/7.

Orphaned Storage

Persistent volumes bound to nothing. Premium storage tiers for workloads that need standard. Disks no pod is reading.

Cost Anomalies

Sudden cost spikes. Resource usage that breaks from its normal pattern. Flagged before the bill arrives.

Six commands. Dashboard included.

Use the CLI for automation and scripting. Use the dashboard for exploration and reporting. Both ship together.

$ kubeopt clusters

List clusters

All connected clusters with health status, monthly cost, and cloud provider.

$ kubeopt analyze prod-aks

Deep analysis

Scans pods, nodes, storage, and billing. Cost breakdown, waste detection, confidence scores.

$ kubeopt report prod-aks

Full report

Per-namespace, per-workload breakdown with every finding and the YAML to fix it. Exportable as JSON.

$ kubeopt compare prod-aks staging-eks

Compare clusters

Side-by-side cost and config diff across any two clusters, including across cloud providers.

$ kubeopt fix prod-aks --type rightsizing

Generate fixes and open PRs (hosted AI)

Reads your infra repo, writes YAML or Terraform patches, opens a pull request. You review and merge.

$ kubeopt chat --cluster prod-aks

Ask questions in plain English (hosted AI)

Get answers grounded in your actual cluster data, not generic advice from documentation.

Commands marked (hosted AI) require a license for the KubeOpt hosted AI service. Everything else runs fully open source, self-hosted.

Open source core. Hosted AI optional.

The cost analysis engine is fully open source. You run it yourself. The hosted AI service is an optional add-on for teams that want automated PR generation and conversational analysis.

Free Open Source Core
  • ✓  Dashboard and cost visualization
  • ✓  Cluster analysis and recommendations
  • ✓  Multi-cloud: AKS, EKS, GKE
  • ✓  Alerts and scheduled analysis
  • ✓  Export reports as JSON
  • ✓  GitHub Action for CI cost gates
  • ✓  MCP server for editor integrations
  • ✓  Self-hosted, your data stays with you
Hosted AI PRO / Enterprise
  • ✓  Everything in the open source core
  • ✓  kubeopt fix: auto-generated YAML and PRs
  • ✓  kubeopt chat: ask questions about your cluster
  • ✓  Implementation plan generation
  • ✓  Priority support

Works where your clusters run

Not "coming soon." Production-tested on all three.

AKS
Azure Kubernetes Service
Azure Cost Management integration. Works with private clusters via Run Command API.
EKS
Elastic Kubernetes Service
AWS Cost Explorer and CloudWatch metrics. STS authentication. Fast local kubectl.
GKE
Google Kubernetes Engine
BigQuery billing integration. Auto-kubeconfig via gcloud. Service account authentication.

Add a cost gate to your CI pipeline

Same idea as a security scan in CI. If your cluster has more than 15% savings available, the pipeline fails and the team knows before the next deploy.

Works in GitHub Actions, GitLab CI, Jenkins, or any pipeline that runs shell commands.

# .github/workflows/cost-check.yml
$ kubeopt analyze prod-aks \
    --ci --threshold 15
Exit 0 = costs within budget
Exit 1 = savings exceed threshold
Read-only access

KubeOpt reads kubectl data and cloud billing APIs. It does not write to your cluster.

You review every change

Fixes are proposed as pull requests. Nothing is applied until you merge.

Self-hosted by default

Runs in your environment. Your cluster data does not leave your network unless you use the hosted AI service.

Try it in two minutes.

No signup. No cloud credentials needed to start. The demo runs on sample data so you can see how it works before connecting anything real.

$ npx kubeopt demo
GitHub Documentation Pricing Live Demo Contact