Fastly Pricing 2026: Plans, Per-TB Costs and Add-Ons
Nov 8, 202614 min read
Fastly's price for a small production app is whatever your bandwidth, request count and add-ons add up to, not the figure on the plan card, which is why it pays to tally your usage, even if you track it with something as simple as a Activepieces workflow that logs the numbers each month, before trusting any headline number.
The right plan is the one where your own monthly traffic makes the next tier cheaper than the usage fees you're paying now. Compare plans by running your numbers through each, and ignore the lowest headline.
Fastly pricing in 2026: the short answer
Fastly is an edge cloud platform
Fastly is an edge cloud platform. It's a network of servers placed close to your users that caches content, delivers it, and runs code and security checks before requests reach your own origin.
For a small app, Fastly bills you for what passes through the network, so a quiet month and a busy month can produce very different invoices on the same plan.
Sources and verification dates for Fastly pricing
I took most figures in this integration from fastly.com/pricing and Fastly's own documentation, and a few from third parties. Where I couldn't verify a number against those sources, the article says so instead of filling the gap with a guess.
If you are running this arithmetic for your own team, see what the same workload costs on Activepieces.
Every current Fastly plan with price and limits
Fastly sells three plans: a free developer tier, pay-as-you-go, and a committed contract. I couldn't verify any dollar minimum or free allowance for them, so each plan below carries a flag where a figure is missing.
Confirm every price on fastly.com/pricing before it reaches your budget.
Fastly free developer tier pricing and limits
This is for developers testing a service or running a hobby project, and it's self-serve. Fastly describes its developer accounts in the post linked above. I couldn't verify the current free allowance, so read it off the pricing page.
Treat it as a capped trial, not a place to run production.
Fastly pay-as-you-go plan pricing
This is the plan for most small production apps. It's self-serve, billed purely on usage, with rates published per unit. The only rates I could source are bandwidth: $0.12 to $0.08 per GB in North America and Europe, covered below, leaving request costs unavailable.
I couldn't verify whether a monthly minimum applies. If one does, it is the real cost for a tiny app.
Fastly committed contract pricing for enterprises
This is for high-volume or security-heavy teams, and it's quote-only. The minimum, allowance and overage rate are set in negotiation, and the contract bundles CDN, WAF, DDoS, compute and support.
No published minimum exists, so I can't give you one, and no break-even calculation involving it is reliable until a quote arrives.
Ask for the minimum commitment, the overage rate and the support terms in writing, then run your own traffic through them. Pay-as-you-go gives you rates to divide by; the committed contract doesn't until someone sends one.
Small production app costs at realistic traffic
Take your own traffic through these terms and the pattern shows quickly: a small production app pays mostly for bandwidth, with request fees a rounding error, and the cache hit rate sets how much traffic reaches your origin.
In a Tencent Cloud example, the cache serves 9,000 of 10,000 requests and 1,000 are origin fetches, so your origin sees one request in ten.
Look at the table below and bandwidth dominates every bill. The table uses round rates of $0.10 per GB and $0.01 per 10,000 requests, resulting costs are approximate rather than exact.
The $0.10 per GB is a placeholder, not a Fastly rate, so actual Fastly costs may be different. The sourced North America and Europe range of $0.12 to $0.08 per GB is the baseline to use.
These are my assumptions, not Fastly's quote, so anyone making a real purchasing decision should verify these figures directly with Fastly. Substitute your quoted rates.
| Delivered | Requests | Bandwidth | Requests cost | Add-ons | Monthly total | Plan |
|---|---|---|---|---|---|---|
| 100 GB | 1M | $10 | $0.10 | $0 | $10.10 | Pay-as-you-go |
| 1 TB | 10M | $100 | $1 | $0 | $101 | Pay-as-you-go |
| 10 TB | 100M | $1,000 | $10 | $0 | $1,010 | Committed contract |
At 100 GB a month, a side project's 1 million requests send 100,000 to your origin. The bill is about $10, so any monthly minimum is the real cost here, not usage.
At 1 TB a month, a small SaaS's ten million requests mean 1 million origin fetches, so 90% of your traffic is served from cache. The bill is about $101.
Bandwidth is 99% of it, so a tier that cuts the per-GB rate matters more than request pricing.
At 10 TB a month the bill is $1,010, and minimums stop mattering: a commitment below that spend costs you nothing extra, and a lower committed rate saves real money.
Usage fees that grow Fastly bills
Once you're past the commitment, the lines that move your invoice are regional bandwidth first, then requests, compute, shielding, image optimisation, security add-ons, log streaming and support. The list below names each metered line, so you can check which ones your traffic will trigger:
- Bandwidth, billed per GB and priced by region
- Requests, billed per 10,000
- Shielding
- Image optimisation
- WAF and bot add-ons
- Compute requests
- Log streaming
- Support
Only bandwidth has rates I can confirm, so it comes first.
Bandwidth by region
Per IO River, North America and Europe run $0.12 to $0.08 per GB, so 1 TB costs $120 at the top rate and $80 at the lowest.
Asia and South America run $0.19 to $0.14, or $190 to $140 per TB, so your geographic footprint significantly affects your bandwidth costs. Korea, India and Africa run $0.28 to $0.24, or $280 to $240 per TB.
At the lowest tiers, a gigabyte served in Korea, India or Africa costs three times one served in North America, so your audience's location can matter more than your volume.
If your app serves many small API responses, you pay mostly on count, not on GB, because Fastly meters Requests and Compute requests apart from bytes. Estimate both numbers.
At the lowest tiers, a gigabyte served in Korea, India or Africa costs three times one served in North America, so your audience's location can matter more than your volume.
Each one you enable adds a charge that your bandwidth figure never shows: WAF, bot protection, shielding, image optimisation and log streaming are each separate lines.
Add support before you compare plans, because Fastly prices it as its own line, separate from usage.
Fastly plan limits and surprises after signup
A plan that looks cheap on your traffic can end up costing more, or stop you shipping, because of any one of these limits. Fastly documents most of them, but they're easy to miss.
A trial is not a free month of production. Fastly's free trial documentation describes restricted account behaviour, and new accounts may need verification before you can serve real traffic.
Read these before you design your setup, since each one caps something you may assume is unlimited:
- Fastly's service limits documentation lists service and configuration limits, such as how many items a single service can hold. Hitting one forces a redesign, not a bigger invoice.
- Purge behaviour differs by method, and purge requests are rate limited. Check the purge documentation if your app invalidates content often, such as per-edit or per-user.
- The API has its own request rate limits, set out in the API reference. Deploy scripts and CI jobs that call it in loops can get throttled.
If your workflow leans on frequent purges or automated deploys, price in the engineering time to batch them.
You won't find overage behaviour and commitment terms on the pricing page; they live in the plan terms and the billing documentation.
Find out whether Fastly bills usage past your allowance at a per-unit rate or blocks it, and whether a commitment locks you in for a term.
Worth checking against a plan that does not meter every step: one credit covers a whole run on Activepieces.
Fastly against Cloudflare, CloudFront and Bunny on cost
Only when your cache hit ratio is high and you'd otherwise buy security separately is Fastly cheaper. The rate card behind a sales call is a cost, so compare plans you can read.
To see which meter your traffic will trip first, use the table, which shows how each vendor bills, not just what it charges (rates as of November 2026).
| CDN | Entry price | Per-GB at 1 TB | Request fees | Included security | Billing model | Beats Fastly when |
|---|---|---|---|---|---|---|
| Fastly | Usage-based, with a free tier | Tiered, falls with volume | Charged per request | Add-on | Metered | Hit ratio is high and you need instant purge |
| Cloudflare | Free plan, then flat monthly | Not metered on plans | Not itemized | Basic DDoS on all plans | Flat plan | Traffic is spiky and you want a fixed bill |
| CloudFront | Pay-as-you-go, with a free tier | Tiered, falls with volume | Charged per request | Add-on | Metered | Your origin already sits in AWS |
| Bunny.net | Pay-as-you-go | Low flat regional rate | Not itemized | Add-on | Metered | Mostly static media, small team |
When Fastly is the cheaper choice
When a request misses the cache, it goes to your origin, and your host bills that transfer. EgressCost lists Google Cloud at $0.12 per GB, AWS at $0.09 and Azure at $0.087.
At 1 TB of misses, that's $120, $90 or $87 on top of the CDN bill. Fastly wins when its caching and purging keep misses low enough that this line stays small.
When a cheaper CDN beats Fastly
If you serve assets from Cloudflare R2, that egress line disappears entirely, since it charges $0 per GB for egress. Pair it with Cloudflare or Bunny and a small app often pays less than on Fastly.
If your origin is already on AWS, CloudFront keeps the transfer inside one bill.
Next Fastly plan upgrade point
Once your own monthly usage, priced at pay-as-you-go rates, costs more than the commitment's minimum, the next plan is cheaper. You can find that point with three invoices and a few lines of arithmetic.
Work through it in this order, using real invoices rather than forecasts:
- Pull your last three invoices and write down the monthly bandwidth in gigabytes and the monthly request count from each.
- Total each month's usage separately, then note the highest and the average. The average tells you what you usually pay; the highest tells you whether one spike is the only reason a commitment looks attractive.
- Price each month at pay-as-you-go using the published per-GB and per-request rates, plus any add-ons you're billed for.
- Price the same usage at the commitment rate on the plan above yours.
- Subtract the plan's minimum from the commitment-rate cost, and treat that minimum as spend you pay even in a quiet month.
- Find the break-even: the monthly usage where pay-as-you-go cost equals the commitment's minimum plus its usage charges. Below it, stay put. Above it, move.
- If your usage lands outside the published rates, stop calculating and ask for a quote.
Check the result against your slowest month, not your busiest. A commitment that wins only in your peak month leaves you paying the minimum for the rest of the year.
When the answer lives only in a quote, go in with your totals and ask specific questions:
- Does unused commitment roll over, or does the minimum expire each month?
- What rate applies to usage above the commitment?
- Which add-ons are bundled, and which are billed on top?
- Can you downgrade at renewal if your traffic falls?
Steps to cap your Fastly bill
Price against your own traffic instead. Pull three months of stats, check your cache hit ratio and regional mix, switch off unused add-ons, and set billing alerts before the next invoice lands.
The steps below turn that history into a decision, and the order matters because each one feeds the next.
- Export three months of bandwidth and request stats from the Fastly dashboard, so you have a baseline to price any tier against.
- Check your cache hit ratio. A low ratio means more requests travel to your origin, so you pay for delivery and for origin load too. Fixing cache rules can cut the bill without touching your plan.
- Break traffic out by region. Delivery rates differ by region, so a mix that leans toward expensive regions can make a higher tier cheaper than your volume suggests.
- List every add-on on the account and switch off any that nobody can say they use. Each one bills whether or not it earns anything.
- Set billing alerts, covered next.
Set a Fastly billing alert early
Set the alert threshold a little under the spend at which the next tier becomes cheaper, using the baseline from step one. Crossing it then tells you to re-run the comparison, not to be surprised at month end.
What Activepieces does about this
The break-even point only helps if someone notices you've crossed it.
In Activepieces you build that check as a scheduled flow: it pulls your Fastly bandwidth and request stats through the HTTP integration, compares them with the threshold you set just under your upgrade point, and posts to Slack or email when usage passes it.
A daily run costs 1 credit, and the free plan's 1,000 monthly credits cover about 30 of those runs. The community edition is MIT-licensed, so you can also self-host it and skip credit counting.
The flow doesn't change what Fastly charges. It tells you when to re-run the comparison.
Frequently asked questions
Does Fastly have a free plan?
Fastly has offered a free tier aimed at developers, but it's a capped trial of the service, not a place to run production indefinitely.
Once your alerts start firing on usage, you're metered like everyone else. Read the current limits on Fastly's pricing page before you build on it.
Is Fastly cheaper than Cloudflare?
It depends on your traffic, and neither headline answers it. Cloudflare leans toward flat plan fees, while Fastly leans toward metering bandwidth and requests.
Light, steady traffic often favors the flat fee. Spiky or heavy traffic is where metered pricing can win or lose, so price both against last month's real numbers.
Does Fastly charge for cache misses or origin traffic?
Fastly bills for what it delivers to your users, and a miss is still a delivered request. The fetch from your origin shows up on your hosting provider's bill, not Fastly's.
Avoid a miss and you save origin load and your host's egress charges. A shield node collapses many edge misses into one origin fetch, so check whether it's billed as an add-on on your plan.
How often does Fastly change its prices?
Fastly revises plans and rate cards from time to time, and the changes don't land on a fixed schedule. Don't budget from a screenshot of last year's page.
Check the current pricing, and compare each invoice line against your own stats. A rate that moved will show up there first.
Related reading
References
Running the numbers
See what the same workload costs here.
Free forever plan, and every paid plan self-hosts at no extra cost.
See pricing Talk to sales
