Estimate Azure Container Apps costs for Consumption and Dedicated profiles with free tier and App Service comparison.
Last verified: May 2026
First 2 million requests per month are free
Consumption plan charges per-second for vCPU and memory used by your replicas, with generous free monthly grants (180,000 vCPU-seconds and 360,000 GiB-seconds). Ideal for bursty, event-driven, or low-traffic workloads that can scale to zero.
Dedicated workload profiles provide reserved compute capacity at fixed hourly rates. Profiles run 24/7 with predictable pricing, similar to VMs. Best for steady-state production workloads requiring guaranteed resources.
Each Azure subscription receives monthly free grants: 180,000 vCPU-seconds (~50 vCPU-hours), 360,000 GiB-seconds (~100 GiB-hours), and the first 2 million HTTP requests. These apply only to the Consumption plan.
Consumption apps can scale to zero when idle, paying nothing. They scale out automatically based on HTTP traffic, KEDA scalers, or custom rules. Dedicated profiles maintain minimum capacity but support higher density and GPU workloads.
Container Apps sits between App Service (fully managed PaaS) and AKS (full Kubernetes). Choose App Service for simple web apps without container orchestration needs. Choose AKS for complex microservice architectures requiring full Kubernetes control. Container Apps offers a serverless container experience with built-in Dapr, KEDA, and Envoy without managing cluster infrastructure.
Output will appear here...Your team migrated a Next.js app from App Service Standard ($73/month) to Container Apps Consumption profile, expecting savings via scale-to-zero. After a month, the bill is $48/month — savings, but less than expected because peak traffic forces 4 active replicas during business hours. The estimator reveals that with the same traffic on Dedicated profile (D4 workload), cost would be $146/month — Consumption is the right call. You also identify a misconfigured min replicas=1 setting wasting $20/month and switch to true scale-to-zero, dropping the bill to $28/month.
The Azure Container Apps Cost Estimator calculates monthly costs for the Container Apps serverless container platform. It covers both Consumption and Dedicated workload profiles, factors in the generous free tier grant (180,000 vCPU-seconds and 360,000 GiB-seconds per subscription per month), and compares costs against equivalent App Service configurations. The tool helps you understand the cost implications of request-driven scaling.
The estimator computes Container Apps cost across three dimensions on Consumption profile: active vCPU-seconds (during request handling), idle vCPU-seconds (between requests if scaled up), and memory GiB-seconds. Each is metered at distinct rates and the per-subscription monthly free tier is subtracted before pricing. For Dedicated profile, it bills on workload profile node-hours like a managed App Service plan.
Container Apps' free tier is per-subscription, not per-app. If you have 10 apps in one subscription, they share 180,000 vCPU-seconds. Spread apps across multiple subscriptions if you need to maximize free tier coverage — but only if the operational complexity is worth the savings.
Scale-to-zero only works on the Consumption profile and only if your app has no minimum replica setting. The default min replicas is 0, but many teams set it to 1 'for warm starts' — which eliminates the scale-to-zero benefit entirely. Cold start in Container Apps is ~5-10 seconds; if your traffic can't tolerate that, use Dedicated profile instead of paying for always-on Consumption.
Container Apps charges for both 'active' and 'idle' time on Consumption. Active = handling requests (full vCPU rate). Idle = scaled-up replica with no traffic (10% rate). This is generous compared to App Service, but means a misconfigured app with 5 always-on replicas idle most of the time still costs real money.
Was this tool helpful?
Disclaimer: This tool runs entirely in your browser. No data is sent to our servers. Always verify outputs before using them in production. AWS, Azure, and GCP are trademarks of their respective owners.