Direct answer: at the Azure Retail Prices API on 9 November 2026 (eastus, USD, excluding tax/discounts), Azure Container Apps Consumption bills “active vCPU-seconds + GiB-seconds + requests”: active vCPU $0.000024/second (about $0.0864/hour, $62.21/vCPU-month), active memory $0.000003/GiB-second (about $0.0108/GiB-hour, $7.78/GiB-month), requests $0.40/million; each month includes 180,000 vCPU-seconds, 360,000 GiB-seconds and 2 million requests free. Idle replicas bill vCPU at $0.000003/second (about $7.78/vCPU-month); memory idle pricing equals active. Environment management/private endpoints/planned maintenance are $0.10/hour (about $73/month). Dedicated bills vCPU $0.057077/hour ($41.67/vCPU-month) + memory $0.004978/GiB-hour ($3.63/GiB-month) + management $0.10/hour ($73/month). For 1 vCPU/2GiB always on: Consumption about $77.76/month, Dedicated about $121.93/month; at 4 vCPU/8GiB, Dedicated about $268.74 vs Consumption about $311.04, with a rough break-even around 3 vCPU/3GiB.

This guide is for Azure application platform, SRE, container/microservice and FinOps teams. It explains Container Apps Consumption/Dedicated billing, active/idle, requests, environment management and free grants; provides 0.5–4 vCPU TCO, Dedicated break-even, the boundary versus Functions/AKS, hidden costs and a 90-day checklist. Prices come from the Azure Container Apps pricing page and Azure Retail Prices API (eastus, 9 November 2026); billing rules/free grants/idle conditions come from Microsoft Learn's Container Apps billing doc. Official price facts, editorial calculations and exclusions are kept separate. This is not procurement advice.

Three rules to remember

(1) Consumption bills active vCPU-seconds + GiB-seconds + requests, and scaling to zero removes compute charges; idle replicas save only vCPU—memory idle pricing equals active. (2) Dedicated bills instance vCPU/memory + management; above roughly 3 vCPU/3GiB and always-on it is cheaper; small scale usually favors Consumption. (3) Environment management/private endpoints/planned maintenance at $0.10/hour, requests, egress, logs, registry and GPU all add cost—do not model vCPU alone.

Plans and billing units: Consumption vs Dedicated

Consumption bills per second: active replicas pay vCPU-seconds + GiB-seconds, HTTP requests per million; scaling to zero incurs no compute charge; with a minimum replica count and idle, vCPU bills at the idle rate while memory stays at the active rate. Dedicated bills by workload-profile instance: vCPU/memory per hour + environment management, offering single tenancy/dedicated hardware/more predictable pricing. Both plans can coexist in one environment; private endpoints/planned maintenance are billed per environment-management hour.

ItemConsumptionDedicated
ComputeActive vCPU-seconds + GiB-secondsInstance vCPU/memory hours
Scale to zeroNo compute chargeInstances always on, billed by allocation
Requests$0.40/million (first 2M free)Not per request (per instance)
ManagementEnvironment $0.10/hour (specific features)$0.10/hour
Best forVariable/low average/event-drivenAlways-on/single tenant/predictable

Source: Azure Container Apps pricing page and billing doc (9 November 2026). Editorial summary; billing follows official pages.

Unit prices: active, idle, requests, environment and Dedicated

Eastus list prices: active vCPU $0.000024/second ($0.0864/hour, $62.21/vCPU-month); active memory $0.000003/GiB-second ($0.0108/GiB-hour, $7.78/GiB-month); requests $0.40/million; idle vCPU $0.000003/second ($7.78/vCPU-month); environment management $0.10/hour ($73/month); Dedicated vCPU $0.057077/hour ($41.67/vCPU-month), memory $0.004978/GiB-hour ($3.63/GiB-month), management $0.10/hour ($73/month); Dynamic Sessions $0.03/hour; Serverless GPU billed per second.

ItemPrice (USD)Monthly/note
Active vCPU$0.000024/secondAbout $62.21/vCPU-month; $0.0864/hour
Active memory$0.000003/GiB-secondAbout $7.78/GiB-month
Requests$0.40/millionFirst 2M/month free
Idle vCPU$0.000003/secondAbout $7.78/vCPU-month; memory same as active
Environment management$0.10/hourAbout $73/month; private endpoint/planned maintenance etc.
Dedicated vCPU$0.057077/hourAbout $41.67/vCPU-month
Dedicated memory$0.004978/GiB-hourAbout $3.63/GiB-month

Sources: Azure Retail Prices API (eastus, 9 November 2026) and Container Apps pricing page. Monthly prices use 730 hours; tax/discounts excluded.

TCO: 0.5–4 vCPU always-on replicas

At a 1 vCPU:2GiB ratio and 730 hours always on: 0.5 vCPU/1GiB Consumption about $38.88/month; 1/2 about $77.76; 2/4 about $155.52; 4/8 about $311.04. Dedicated adds $73/month management: 0.5/1 about $97.46; 1/2 about $121.93; 2/4 about $170.87; 4/8 about $268.74. Dedicated becomes cheaper above roughly 3 vCPU/3GiB when always on/long term.

Size (always on)ConsumptionDedicated
0.5 vCPU/1GiB38.8897.46
1 vCPU/2GiB77.76121.93
2 vCPU/4GiB155.52170.87
4 vCPU/8GiB311.04268.74

Editorial calculation: active rate ×2,592,000 seconds; Dedicated adds $0.10/hour management; requests/environment/egress/logs/tax/discounts excluded; actual vCPU:GiB ratios vary.

Break-even: Dedicated, idle, requests

Dedicated vs Consumption: at 1 vCPU:1GiB, Consumption is about $69.98/vCPU-month, Dedicated about $45.30/vCPU-month + $73 management → a break-even around 3 vCPU/3GiB; at 1:2, Consumption about $77.76/vCPU-month, Dedicated about $48.93/vCPU-month + $73 → about 2.5 vCPU/5GiB. Idle: 1 vCPU idle $7.78/month vs active $62.21; memory does not save. Requests: 2M free; 100M about $39.20; 1B about $399.20.

ComparisonCalculationConclusion
Dedicated vs Consumption(41.67+3.63) + 73 vs (62.21+7.78)Above about 3 vCPU/3GiB Dedicated is cheaper (always on)
Idle vs active vCPU7.78 vs 62.21/vCPU-monthIdle saves vCPU only; memory same
Requests 100M(100M−2M)×$0.40/MAbout $39.20
Requests 1B(1B−2M)×$0.40/MAbout $399.20
Scale to zeroNo compute chargeBest for event-driven/variable

Editorial calculation: 730 hours/month, 1:1 vCPU:GiB; Dedicated management $73/month; environment/egress/logs/tax/discounts excluded; actual ratios/discounts change the threshold.

Hidden costs: seven items beyond vCPU

Common add-ons: environment management/private endpoints/planned maintenance ($0.10/hour, about $73/month), requests ($0.40/million, first 2M free), idle replicas (vCPU $7.78/month; memory not saved), egress (cross-region/public), logs/OpenTelemetry (ingestion/retention; the managed agent itself has no extra compute charge), container registry/storage (ACR, volumes), Serverless GPU (T4/A100 per second) and Jobs/Dynamic Sessions (execution/session hours).

Cost itemTriggerControl
Environment managementPrivate endpoint/planned maintenance etc.Enable as needed
Requests>2M/monthCache/coalesce/static content
Idle replicasMin replicas >0Scale to zero/keep on demand
EgressCross-region/publicSame region, compression
Logs/OTelIngestion/retentionSampling/retention
ACR/storageImages/volumesCleanup/lifecycle
GPU/JobsGPU/batchOn demand/time-boxed

Sources: Container Apps pricing/billing docs (9 November 2026). Editorial summary; billing follows official pages and bills.

How to choose: Container Apps vs Functions vs AKS

Containerized/microservices/long-running HTTP with scale-to-zero or per-request billing → Container Apps Consumption; always-on/single tenant/dedicated hardware/predictable → Dedicated; event-driven small functions/short tasks billed per execution → Azure Functions Flex; full Kubernetes API/ecosystem/custom controllers → AKS (control-plane costs separate). They can combine: Container Apps for APIs/microservices, Functions for events, AKS for complex platforms.

ScenarioPreferCheck first
Containerized API/microservices/variableContainer Apps ConsumptionvCPU/GiB, requests, scale to zero
Always-on/single tenant/dedicated hardwareDedicatedInstance size, management fee
Events/short tasks/small functionsAzure Functions FlexExecutions/GB-seconds, cold start
Full K8s/custom controllersAKSControl plane/nodes/operations
Batch/scheduledContainer Apps JobsExecution duration/retries
GPU inferenceServerless GPUGPU-second price, cold start

Editorial selection framework, not a vendor recommendation; features/compliance/regional availability follow official docs and business requirements.

90-day checklist

Use the first 90 days to close “replica/request baseline → plan → pilot → benchmark → cutover → review.” Weeks 1-2: pull replica counts, vCPU/GiB, requests, scale-to-zero duration and min replicas; weeks 2-4: calculate Consumption/Dedicated break-even; weeks 4-6: pilot and test cold start/concurrency/scale-down; weeks 6-8: configure environment/private endpoint/logs; weeks 8-10: cut over and monitor; weeks 10-12: cost review, tune min replicas/sizes. Do not keep a large min replica “for a possible peak.”

FAQ

How is Container Apps billed?

Consumption bills active vCPU-seconds ($0.000024) + GiB-seconds ($0.000003) + requests ($0.40/million); free 180k vCPU-s, 360k GiB-s, 2M requests. Dedicated bills vCPU ($0.057077/hour) + memory ($0.004978/GiB-hour) + management $0.10/hour.

How much is 1 vCPU/2GiB always on?

Consumption about $77.76/month; Dedicated about $121.93/month (including $73 management). 4 vCPU/8GiB: Consumption about $311.04, Dedicated about $268.74.

When is Dedicated cheaper?

When always on and above roughly 3 vCPU/3GiB, Dedicated's lower vCPU/memory rates amortize the management fee; small/variable workloads favor Consumption.

Does scale-to-zero still cost?

No compute charge; requests/egress/logs/storage may still apply. With min replicas, idle vCPU is $7.78/vCPU-month and memory stays at the active rate.

How are requests billed?

$0.40/million, first 2M/month free; only external HTTP requests into the environment are billed; health probes are not, and Jobs do not bill requests.

Container Apps or Functions?

Containers/long-running HTTP/microservices → Container Apps; event-driven/short tasks/small functions → Functions Flex. Both are usage-based but use different units (vCPU/GiB-s vs executions/GB-s).

Container Apps or AKS?

Full Kubernetes API/custom controllers/ecosystem → AKS; just run containers/microservices with less cluster management → Container Apps. AKS also has control-plane/node/operations costs.

When is environment management charged?

Private endpoints, planned maintenance and similar features bill $0.10/hour (about $73/month); the default environment may not always charge—check the official billing doc.

Research method and limitations

This article separates “official price facts” from “editorial calculations.” Consumption active/idle vCPU, memory, requests, environment management and Dedicated vCPU/memory/management come from the Azure Container Apps pricing page and Azure Retail Prices API (eastus, 9 November 2026); free grants, active/idle conditions, request/Jobs/environment rules come from Microsoft Learn's Container Apps billing doc. TCO, break-even, hidden costs and the 90-day checklist are editorial.

Eight limitations: (1) only eastus list prices, excluding tax/discounts/enterprise agreements; (2) 730 hours and 1:1 or 1:2 vCPU:GiB assumptions; (3) requests/environment/egress/logs/ACR/storage/GPU combinations are excluded; (4) no real cold-start/concurrency/scale-down benchmark; (5) Dedicated sizes/availability/management fees vary by region/environment; (6) Container Apps Jobs/Dynamic Sessions/Serverless GPU billing is not fully covered; (7) savings plan 1/3-year discounts are not modeled; (8) prices/features change, so recheck official pages and bills before purchase.

Sources and basis

Data cut-off 9 November 2026; prices/availability follow the Azure official page and bill; this is not procurement, commercial or legal advice.

Turn replicas/requests, Consumption/Dedicated and hidden items into a checkable budget

Bring replica/vCPU/GiB, requests, scale-to-zero, min replicas and environment/private-endpoint needs; we turn Container Apps cost into a checkable budget.

Book a Container Apps cost review