Cloudinary Pricing in 2026: What You Actually Pay

Cloudinary bills on a currency it invented: the credit. That single design choice is why so many teams read the pricing page, pick a plan, and still get a surprise 30 days later. This teardown converts credits back into dollars and shows you, with real 30-day math, which line on your usage graph actually blows the budget.

Cloudinary is a media platform: it stores your images and video, transforms them on the fly (resize, crop, reformat, compress), and delivers them over a CDN. You pay for all three through one shared unit.

What one Cloudinary credit actually buys

Here is the rule that everything else depends on. Per Cloudinary's own credits documentation (2026), one credit equals any one of these:

Credits are fungible. Your monthly allotment is one pool, and storage, transformations, and delivery all draw from the same bucket. That is the part people miss: you do not get "225 GB of storage and 225,000 transformations." You get 225 credits, total, and whichever activity is heaviest eats the pool first.

The plans, in dollars (2026)

Cloudinary's self-service ladder has three public rungs plus a custom tier. Prices from the Cloudinary pricing page, July 2026:

Two things jump out of that last column. First, paying annually knocks roughly 10% off. Second, the effective price per credit barely improves as you climb: about $0.40 on Plus, about $0.37 on Advanced. There is no volume discount worth chasing on the self-service tiers. You are paying roughly the same unit rate; you are just buying a bigger bucket.

Because 1 credit also equals 1 GB of delivery, that unit rate is the number to remember: you are paying roughly $0.37 to $0.44 for every GB Cloudinary delivers, once you are on a paid plan. Hold that thought.

Which lever runs out first (three real 30-day bills)

Sticker price tells you nothing until you model usage. So let us model three real shops. Assumptions, stated up front so you can argue with them:

• Average stored original: 1.5 MB
• Average delivered image after f_auto and q_auto (Cloudinary's automatic format and quality): 120 KB
• 1 credit = 1 GB storage = 1 GB delivery = 1,000 transformations

These are BudgetForge estimates, not Cloudinary quotes. Your mileage varies, but the shape holds.

Cohort A: the weekend project

• Storage: 2,000 originals x 1.5 MB = 3 GB, so 3 credits
• Transformations: ~8,000 / mo, so 8 credits
• Delivery: 40,000 image loads x 120 KB = ~4.6 GB, so ~5 credits
• Total: ~16 credits

Sixteen credits sits comfortably inside the Free tier's 25. Your bill is $0, and it should stay there for a while. If you are a solo builder shipping an MVP, Cloudinary's free plan is genuinely generous. Take it.

Cohort B: the growing marketplace

• Storage: 20,000 originals x 1.5 MB = 30 GB, so 30 credits
• Transformations: ~40,000 / mo, so 40 credits
• Delivery: 500,000 image loads x 120 KB = ~57 GB, so ~57 credits
• Total: ~127 credits

You have blown past Free and need Plus ($89 to $99/mo) with its 225 credits. You have headroom, but look at the mix: delivery (57) is already your single biggest line, bigger than storage and transformations combined. That is the pattern. Delivery scales with pageviews, and pageviews grow faster than your image library does.

Cohort C: the image-heavy store having a good month

• Storage: 60,000 originals x 1.5 MB = 90 GB, so 90 credits
• Transformations: ~150,000 / mo, so 150 credits
• Delivery: 3,000,000 image loads x 130 KB = ~372 GB, so ~372 credits
• Total: ~612 credits

Now it hurts. 612 credits is past Plus (225) and past Advanced (600). You are in Enterprise-quote territory. And notice: delivery alone (372 credits) is larger than the entire Advanced plan. A traffic spike, not a wave of new uploads, is what pushed you over. The thing you were celebrating (a good sales month) is the thing that broke the budget.

The rolling-window trap

Here is a detail almost every listicle skips, and it has cost people real money. Cloudinary meters usage over a rolling last-30-days window, not a calendar month that zeroes out on the 1st. Cloudinary's support team confirms it: your usage is measured across "a rolling last-30-days time window."

Practical translation: if you have a viral Tuesday, that spike keeps counting against your allowance for the next 30 days. It does not evaporate when the calendar flips. On the self-service plans, Cloudinary generally does not quietly bill you a smooth per-GB overage either; it prompts you to upgrade to the next tier or a custom plan. So the jump is a cliff, not a ramp: you go from $89 to $249, or into a sales call, with little in between.

If your traffic is spiky, budget against your worst rolling 30 days, not your average month.

The cheaper-lever move

Cloudinary's transformation engine is genuinely good, and that is what you are actually paying for. The mistake is paying its metered rate for the two things it is worst-value at: cold storage and repeat delivery. Three moves, cheapest first.

1. Turn on f_auto and q_auto. Automatic format (serve AVIF or WebP where supported) and automatic quality routinely cut delivered bytes 30% to 70%. Because delivery is billed by the GB, cutting bytes cuts credits one for one. This lever is free and it is the highest-ROI thing on this list.

2. Cache transformed URLs at your own edge. Cloudinary bills net viewing bandwidth, the bytes it serves. Put a CDN you control in front (many teams use a plain caching layer), and repeat hits on the same transformed URL get served from your cache without touching Cloudinary's meter. At Cohort C, that turns ~372 delivery credits into a fraction of that.

3. Keep cold originals in cheap object storage. At roughly $0.37 to $0.44 per GB, Cloudinary is an expensive place to warehouse originals you rarely transform. Object storage is 15x to 25x cheaper per GB. We ran the numbers on the cheapest option in a separate teardown; see cloudflare-r2-pricing-2026 for R2's rates and the zero-egress angle. Store originals there, pull into Cloudinary only what you actively transform.

Do all three and you keep Cloudinary for the one job it is worth paying a premium for, the transformation pipeline, while paying commodity rates for storage and delivery.

How it compares on raw delivery

To be fair, Cloudinary is not sold as a bare CDN, and comparing it to one is apples-to-oranges on features. But on the specific line item of delivery bandwidth, the gap is worth seeing.

We priced the usual alternatives people reach for: Bunny.net, imgix, Cloudflare Images, ImageKit, and AWS CloudFront. Bunny.net's CDN lists around $0.01 per GB in North America and Europe. AWS CloudFront lists roughly $0.085 per GB on its first tier. Cloudinary's blended delivery works out to about $0.37 to $0.44 per GB.

That is not a knock on Cloudinary; it is a knock on using Cloudinary as your delivery CDN at scale. The transformations are the value. The bytes are not.

So who should pay for Cloudinary?

• Free tier: hobby projects, MVPs, and small sites under ~25 credits of real usage. Excellent value at $0.
• Plus ($89 to $99): growing apps that lean on Cloudinary's transformation and DAM features and land comfortably under 225 credits with the three cost levers applied.
• Advanced ($224 to $249): teams using video processing, more environments, or heavier transformation volume who still fit under 600 credits.
• Enterprise: high-delivery businesses, but only after you have offloaded repeat delivery and cold storage. If delivery is still your biggest credit line when you call sales, you are negotiating from the wrong position.

If your usage is dominated by delivery bandwidth and you are not using much of the transformation engine, Cloudinary is probably the wrong tool, and a CDN plus a lighter optimizer will cost a fraction.

Math check: on a paid plan you pay about $0.40 per credit, and 1 credit equals 1 GB delivered, so delivery bandwidth is roughly $0.40/GB, 4x to 40x a raw CDN. In every cohort past the free tier, delivery, not storage, is the credit that runs out first.