A 50 TB/month video workload split 60/40 between North America and Asia Pacific generates an AWS CloudFront bill roughly 35% higher than the same bytes delivered from a single US-centric region. That delta — invisible if you only glance at the headline $0.085/GB rate — is where CloudFront pricing actually lives. This article gives you the full framework: region-by-region rates as of Q2 2026, tier arithmetic for real traffic shapes, the cost traps that inflate bills at scale, a workload-profile decision matrix you won't find in AWS docs, and concrete dollar examples from 1 TB to 500 TB. If you forecast CDN spend for a living, this is the reference you keep open in a tab.
AWS still runs two billing tracks for CloudFront: classic pay-as-you-go and the Security Savings Bundle introduced in late 2023 (which bundles CloudFront + WAF spend for a 1-year commitment at up to 30% savings). As of April 2026, no new pricing tiers have been added to the pay-as-you-go schedule. The key structural detail that matters most: data transfer from AWS origins (S3, ALB, API Gateway, MediaStore) to CloudFront edge locations carries zero transfer cost when the response is served through a CloudFront distribution. That zero-cost origin fetch is the single biggest reason AWS-native architectures look attractive on a CloudFront bill — and the single biggest reason those bills become harder to compare against CDNs that charge a flat per-GB rate regardless of origin location.
Two smaller 2026 updates worth noting. First, CloudFront Functions pricing remains at $0.10 per million invocations with no time limit, while Lambda@Edge still bills per request plus per GB-second of compute. Second, real-time log delivery to Kinesis Data Streams is priced at $0.01 per million log lines — a cost that surprises teams running high-request-rate API distributions for the first time.
The table below reflects AWS's current published pay-as-you-go data transfer out (DTO) pricing for CloudFront. All prices are per GB. Tiers are cumulative per month, across all distributions on the account.
| Monthly Tier | US/Canada/EU | Japan/HK/PH/SG/TW/KR | Australia/NZ | South America | India |
|---|---|---|---|---|---|
| First 10 TB | $0.085 | $0.114 | $0.114 | $0.110 | $0.109 |
| Next 40 TB | $0.080 | $0.089 | $0.098 | $0.090 | $0.085 |
| Next 100 TB | $0.060 | $0.075 | $0.075 | $0.080 | $0.070 |
| Next 350 TB | $0.040 | $0.060 | $0.060 | $0.060 | $0.056 |
Request charges sit on top. HTTPS GET requests cost $0.0100 per 10,000 in the US/EU and $0.0125 per 10,000 in most APAC regions. HTTP requests are roughly half that. For a workload serving 500-byte API responses at 100 million requests/month in the US, the request line item alone is $100 — potentially more than the DTO if your total transfer is under 2 TB.
Price classes let you exclude expensive edge locations. Price Class 100 restricts delivery to US, Canada, and Europe. Price Class 200 adds Japan, Hong Kong, Singapore, and a few other APAC locations. Price Class All uses every edge location globally. The cost difference is material: a workload that would incur $0.114/GB from Australian edges costs $0.085/GB if those viewers fall back to a US edge via Price Class 100. The tradeoff is latency. For static asset delivery where TTFB matters less than total cost, restricting price classes is a valid optimization. For live video or real-time APIs serving APAC users, the latency penalty usually outweighs the savings.
Transfer: 4.5 TB × $0.085 (US) + 0.5 TB × $0.114 (APAC) = $382.50 + $57.00 = $439.50. HTTPS requests at 20 million: $20.00. Total: approximately $460/month. Effective rate: $0.090/GB.
US/EU transfer (first 10 TB at $0.085, next 35 TB at $0.080): $850 + $2,800 = $3,650. India transfer (5 TB at $0.109): $545. Requests at 5 million (large objects): $5. Total: approximately $4,200/month. Effective rate: $0.084/GB blended.
Tier math across US/EU: first 10 TB at $0.085 + 40 TB at $0.080 + 100 TB at $0.060 + 50 TB at $0.040 = $850 + $3,200 + $6,000 + $2,000 = $12,050. This excludes APAC and South America traffic, which at even 15% of volume adds $1,500–$2,000. Requests on large binaries are minimal. Total estimate: $14,000–$14,500/month. Effective blended rate: approximately $0.072/GB.
At this scale, the economics shift considerably. For enterprises delivering 200 TB+ monthly, alternatives with flat per-TB pricing become worth modeling. BlazingCDN delivers comparable stability and fault tolerance to CloudFront at a fraction of the cost: $1,500/month covers up to 500 TB (effective rate of $0.003/GB), scaling down to $0.002/GB at 2 PB. For a 200 TB workload, that is a difference of roughly $12,000/month versus CloudFront's $14,000+ — with simpler billing and no region-based rate variance. Sony is among BlazingCDN's clients operating at this tier.
This matrix is the section you won't find in AWS docs or competing blog posts. It maps workload profiles to the pricing model that actually minimizes total cost of delivery.
| Workload Profile | Monthly Volume | CloudFront Fit | Key Consideration |
|---|---|---|---|
| AWS-native app (S3 + ALB origin) | Under 50 TB | Strong | Zero origin-to-edge transfer cost is a real advantage |
| API / microservice gateway | Any | Moderate | Request charges dominate; model per-request cost carefully |
| VOD / large-file download | 100 TB+ | Weak on cost | Flat-rate CDNs at $0.003–$0.004/GB save 50%+ at this scale |
| Live streaming, APAC-heavy audience | 50+ TB | Expensive | APAC rates at $0.114/GB make multi-CDN strategy essential |
| Software/game patch distribution | 200 TB+ | Weak on cost | Large objects + global audience = evaluate dedicated delivery CDN |
| Multi-origin, non-AWS backend | Any | Neutral | Origin transfer waiver doesn't apply; no inherent cost advantage |
Serving HLS manifests, API responses, or thumbnails generates high request counts relative to bytes transferred. A distribution serving 200 million HTTPS requests/month in the US adds $200 in request charges alone. If average response size is 2 KB, total transfer is only 400 GB ($34 in DTO). Your effective per-GB rate is $0.585/GB. The AWS CloudFront pricing calculator won't flag this unless you input request counts explicitly.
The first 1,000 invalidation paths per month are free. Each additional path costs $0.005. A CI/CD pipeline that purges 50,000 paths per deployment will spend $245 per purge event. This is almost always a cache-key design problem. Versioned URLs or cache-tag-based purging (via CloudFront Functions rewriting cache keys) eliminates this entirely.
A deployment that changes query string parameters, Vary headers, or cookie forwarding rules can shatter cache partitions. Cache-hit ratio drops from 95% to 40%, origin load quadruples, and your origin transfer bill — if on a non-AWS backend — spikes. Monitor cache-hit ratio in CloudFront's built-in metrics within 30 minutes of every deploy.
At $0.01 per million log lines, a 100 million request/month distribution generates $1/month in log costs. At 10 billion requests/month, that's $100 — not trivial, and easy to overlook when you're focused on DTO.
In the US, Canada, and Europe, CloudFront charges $0.085/GB for the first 10 TB/month, dropping to $0.080 for the next 40 TB and $0.060 for the next 100 TB as of Q2 2026. APAC regions start at $0.114/GB. Your effective per-GB rate depends on region mix and volume tier.
Yes. CloudFront's always-free tier includes 1 TB of data transfer out, 10 million HTTP/HTTPS requests, and 2 million CloudFront Functions invocations per month. This applies indefinitely, not just for the first 12 months. It is sufficient for low-traffic sites but irrelevant for production workloads at scale.
Price Class 100 limits delivery to US, Canada, and European edges, avoiding the $0.114/GB APAC rate. Price Class 200 adds select APAC locations. Viewers in excluded regions are served from the nearest included edge, increasing latency but reducing your DTO rate. This is a valid optimization for cost-sensitive static asset delivery.
Combine the Security Savings Bundle (up to 30% commitment discount), restrict to Price Class 100 if your audience allows it, maximize cache-hit ratio above 95%, and use versioned URLs to eliminate invalidation costs. Even with all optimizations, CloudFront at 100 TB/month in the US costs roughly $6,850/month in DTO alone.
Requests are billed separately from data transfer. HTTPS requests cost $0.0100 per 10,000 in US/EU and $0.0125 in APAC. HTTP requests cost roughly half. For workloads with small response sizes and high request rates, request charges can exceed data transfer costs.
Data transfer from supported AWS origins (S3, ALB, MediaStore, API Gateway) to CloudFront is free when served via a CloudFront distribution. Data transfer from non-AWS origins is billed at standard internet transfer rates from the origin's location, which is a separate line item from CloudFront DTO.
Pull your last 90 days of CloudFront usage from Cost Explorer. Split by region and request class. Calculate your actual effective per-GB rate — not the headline number, the one that includes requests, invalidations, and real-time logs. Compare that number against flat-rate alternatives at your volume tier. If the gap exceeds 30%, you have a procurement conversation worth having before your next annual commit. Post your blended rate (anonymized) in your team's infra channel and see if anyone is surprised. They usually are.