Cloudflare R2 2026 Pricing Features and AWS S3 Comparison

When I first started using cloud storage for my projects, I didn’t pay much attention to egress fees. I assumed they were just part of running anything online. But the deeper I went into hosting media files and high-traffic assets, the more I realized how destructive these fees can be. AWS, for example, charges $0.09 per GB of data transferred to the internet. That means a simple site serving around 20 TB every month ends up paying roughly $1,700 just for bandwidth. It shocked me how quickly that adds up, especially when traffic scales faster than revenue.

That’s what led me to explore Cloudflare R2. The idea of zero egress fees sounded almost unreal at first. But once I started comparing the costs and testing the workflows myself, I noticed how dramatically different my bills became. Suddenly, the storage bill was predictable again. No more surprises. And for the first time, I felt like I wasn’t being penalized for growing my audience.

I wanted to break down my experience in a clearer, more personal way, especially as R2’s pricing structure moves into 2026. I also wanted to share what I learned about where R2 excels, where AWS S3 still makes sense, and how I now decide between the two depending on the workload.

Understanding Cloudflare R2 from My Perspective

I remember when R2 launched in 2021. Cloudflare positioned it as a direct challenge to AWS S3, and that caught my attention because S3 had long been the industry default. What truly separated R2 wasn’t the storage price or the API compatibility—it was the decision to eliminate egress fees entirely.

For me, that single change shifted the entire cost equation. R2 integrates with Cloudflare’s global network of over 300 data centers, which means my stored content is automatically served from the closest edge location. I don’t have to configure regions or think about replication rules; R2 handles it behind the scenes.

Another thing I appreciated was the S3 compatibility. I could point my existing tools and code toward R2 simply by switching the endpoint URL. I didn’t have to rewrite anything. SDKs, libraries, workflows—they all behaved the same way.

How R2 Pricing Works for Me

Heading into 2026, R2’s pricing has stayed simple. I pay for storage, Class A operations, and Class B operations. Storage costs $0.015 per GB per month. If I store 1 TB, I pay $15. That’s it.

Cloudflare bills storage based on average daily usage. So if I stored 100 GB for half the month and 200 GB for the other half, I’m billed as if I stored 150 GB. It’s straightforward and easy to predict.

Class A operations—uploads, deletes, and listing requests—cost $4.50 per million, and the first million every month is free. Most of my projects barely scratch that limit.

Class B operations—essentially reads—cost $0.36 per million, with the first 10 million free. For context, a site with 300,000 page views where each page loads three images would still stay within the free tier.

Cloudflare also introduced an Infrequent Access tier a while ago. It costs less for storage but charges more for operations and has a 30-day minimum usage requirement. I use this tier only for data I know I won’t delete or modify for at least a month.

What surprised me the most, though, was how generous the free tier is. With 10 GB of storage, one million Class A ops, and ten million Class B ops every month, I realized I could run small sites at zero cost.

Comparing This to AWS S3

I’ve used AWS S3 for years, and its pricing structure is powerful but complex. S3 Standard starts at $0.023 per GB in most US regions. There are multiple storage classes—Standard, Standard-IA, One Zone-IA, Glacier, Glacier Deep Archive—and the costs vary across regions.

Request pricing is also layered: PUT requests cost around $0.005 per thousand, while GET requests cost around $0.0004 per thousand. It seems small at first, but with large workloads, it grows quickly.

The real challenge, though, is data transfer. The first 100 GB out of S3 is free every month, but after that, I pay $0.09 per GB for the first 10 TB. When I calculated costs for a media project serving 20 TB monthly, I realized egress alone costs over $1,700.

Cloudflare R2, by comparison, would cost me just $15 for that same workload, because the bandwidth is free.

When R2 Saves Me Money

If my workload involves delivering large amounts of content—images, video, software downloads—R2 wins every time. The cost advantage is huge. The more bandwidth I serve, the more dramatic the savings become.

For example, storing 1 TB and serving 20 TB monthly on S3 costs about $1,723. R2 costs $15. The difference is unbelievable.

If storage retrieval is frequent or unpredictable, R2’s pricing is also cleaner and easier to estimate. I don’t have to think about cross-region requests, replication fees, or egress multipliers.

When S3 Still Makes Sense for Me

Even though I love the pricing model of R2, there are situations where AWS S3 remains the better choice.

If I’m storing hundreds of terabytes of archival data, Glacier Deep Archive is extremely cheap. R2’s Infrequent Access tier helps, but it can’t compete with Glacier for deep storage.

When I’m building projects that depend heavily on the AWS ecosystem—Lambda triggers, Athena queries, Glue pipelines—S3 is the natural fit. AWS has nearly two decades of maturity and integrations.

S3 also offers advanced compliance features like Object Lock and Access Points. If a project requires strict governance and audit trails, S3 handles that more comprehensively.

What I Like About R2 Beyond Pricing

The simplicity is refreshing. I don’t pick regions or deal with replication rules. Everything is automatically available across Cloudflare’s network. That alone saves me time.

I also use Cloudflare Workers often, and R2 integrates seamlessly with it. Having compute and storage share the same edge network speeds up a lot of my workflows.

Recently, event notifications were introduced, letting me trigger Workers when objects change. That brings R2 much closer to S3’s event-driven capabilities.

Performance Impressions

Cloudflare’s global edge network gives R2 a speed advantage for public assets. Their tests showed 20–40% faster response times than S3 for public bucket access. I’ve seen similar improvements in my own projects, especially for users outside my primary region.

My Migration Experience

AWS removed egress fees for migrations in 2024, which made moving my content much easier. I used Cloudflare’s Super Slurper tool to copy buckets over. It handled everything and only billed me for the operations on R2’s side.

For ongoing migrations, I used Sippy, which pulls objects into R2 as they’re requested. That allowed a smooth, zero-downtime transition.

How I Choose Between R2 and S3 Today

If my workload involves heavy data delivery, public media, or anything bandwidth-intensive, I choose R2. The savings aren’t small—they’re transformative.

If the workload is deeply integrated into AWS services or compliance requirements, I choose S3.

Sometimes, I even combine both. R2 for high-traffic public assets, S3 for deep archives or AWS-native pipelines. It gives me the best balance of cost and capability.

Final Thoughts

As 2026 approaches, the divide between S3 and R2 has become clearer for me. S3 remains unmatched for enterprise features, compliance, and ecosystem depth. But R2 is a game-changer for bandwidth-heavy use cases. Its zero-egress model feels like the future of object storage.

Whenever I plan a new project now, I always calculate the total cost, especially the bandwidth portion. And more often than not, R2 ends up being the smarter, more predictable, and far more affordable option.