Bunny Stream vs FastPix: why a per-gigabyte video bill grows faster than your audience

You can just put your video on Bunny Stream. It is a good product. The transcoding is free, the player is free, multi-DRM (digital rights management) is included, and one cent per gigabyte (GB) is a real number on a real pricing page. Plenty of teams should buy it, and if you are one of them we would rather you knew that from us than found out in month four.

But before assuming it is the cheaper option, it is worth doing the comparison properly, because the two pricing models behave in completely different ways as you grow.

Is Bunny Stream cheaper than FastPix?

Bunny Stream charges for gigabytes delivered. FastPix charges for minutes delivered. That single difference is the whole comparison, and two things move it.

The first is bitrate. A gigabyte bill goes up every time your video gets sharper. A per-minute bill does not move at all.

The second is geography, and this is the one most teams miss. Bunny CDN pricing is regional, so the rate is not one cent per gigabyte everywhere. One cent is the Europe and North America rate. Asia and Oceania is three cents, South America is 4.5 cents, and the Middle East and Africa is six cents. Bunny publishes all four, so none of this is hidden. It just means the number you budget from is the cheapest of four, and you only pay it if every viewer you have lives in one of two regions.

FastPix charges the same rate in every country on earth.

How does Bunny Stream pricing actually work?

Everything below comes out of one conversion, so it is worth doing once by hand.

1 minute at b Mbps  =  60 × b megabits
                    =  7.5 × b megabytes
                    =  0.0075 × b gigabytes

1080p at 5 Mbps     =  0.0375 GB per minute
  at $0.01/GB       =  $0.000375 per minute
  at $0.03/GB       =  $0.001125 per minute

FastPix's Standard media quality tier, which encodes free, bills that same 1080p minute at $0.000679 wherever it lands. Below 1080p, every rendition from 240p up to 720p bills at $0.000543. Streaming is billed on the rendition actually delivered, not on the resolution the asset was encoded at, so a viewer who pulls 480p from a 1080p master is billed at the lower rate.

Bunny's free encoding tier produces a fixed ladder in H.264, the codec every phone decodes in hardware. These are the bitrates that go into the gigabyte side of the maths:

Bunny's documentation notes bitrate can run up to 1.4 times these defaults depending on the source file, which moves the gigabyte bill up and never down.

Are the two ladders the same quality?

A bitrate table invites you to assume that 2,800 kbps is 2,800 kbps. Three things it does not show change both what the viewer sees and what the bytes cost.

Frame rate. Bunny's ladder is typically delivered at 30 frames per second. FastPix encodes every rendition at 60, so motion stays smooth from the bottom of the ladder to the top rather than only on the rungs a viewer may never reach.

How the bits are assigned. Bunny's numbers are fixed defaults, scaled up to 1.4 times for a demanding source. FastPix's 1080p bitrates are content-aware: the encoder analyses the actual footage and spends bits where the picture needs them, which holds quality on hard scenes without wasting bandwidth on the easy ones.

Audio, which is where this becomes a pricing question. Bunny bundles audio into the video at a single fixed quality, so every viewer receives the same audio track no matter which video rendition they are served. On a per-gigabyte bill that lands hardest on exactly the viewers who should be cheapest: someone on the 240p rung is pulling 600 kbps of video and carrying a full-quality audio track on top of it, and you pay for those bytes. FastPix serves audio as three separate quality tracks tied to the video rendition, so a viewer on a low rung saves bandwidth on both halves of the stream.

There is a second-order point worth stating plainly. On a per-gigabyte bill, every encoding improvement that raises quality also raises your invoice, so quality and cost pull against each other. On a per-minute bill they do not. A better encode costs you nothing, because you are not paying for the bytes.

Bunny CDN pricing is regional, and one cent is the best case

Take Bunny's four published rates and blend them across an actual audience. Four profiles, each stated as an assumption rather than a claim, so you can substitute your own:

A globally distributed audience pays about 3.3 cents a gigabyte, not one. Bunny's one cent is not a price. It is a best case that survives exactly as long as your audience stays in two regions.

What a million minutes costs

One 1080p asset at 5 Mbps, a million minutes delivered a month. Assumptions stated, arithmetic shown:

Bunny   = 1,000,000 min × 5 Mbps × 0.0075 GB × blended rate
FastPix = 1,000,000 min × $0.000679

Three of four go one way and the fourth goes the other, which is roughly what an honest comparison looks like. If your entire audience is in the United States and Europe and your content is not especially sharp, Bunny is cheaper today and you should buy Bunny.

The word doing the work in that sentence is today.

What happens when it works

Here is the same product, four years apart. Same content, same bitrate, same everything except that it succeeded and the audience spread out.

Read the last row against the first. The audience grew twelve times. The FastPix bill grew twelve times, because that is the only thing it is a function of. The Bunny bill grew almost forty times.

Nothing went wrong in that table. No rate changed, no vendor raised a price, nobody made a bad decision. The company simply got what it wanted, and the pricing model charged it a second time for the getting. On a per-gigabyte bill, success is the expensive part.

That is the difference worth understanding before you pick. Bunny charges you for where your viewers live. FastPix charges you for how long they watched. Only one of those is a number your product team controls.

Where does your product land?

Bitrate and audience map are both properties of what you are building, so different products sit in different places. A million minutes a month in each:

Education is the interesting row. Learning platforms tend to serve a concentrated regional audience at a moderate bitrate, which is precisely the shape per-gigabyte pricing is good at, and Bunny wins it on delivery. That row does not change when we would like it to.

Microdrama is the other one worth reading carefully. Short vertical episodes have a low blended bitrate, because a ninety-second episode spends much of its runtime in the adaptive bitrate ladder's ramp rather than at the top of it. The margin there is seven percent, which is close enough to be decided by your actual rendition mix rather than by anything in this table.

What about Bunny's Volume network?

This is the right question, and the real answer is that Bunny's Volume network is cheaper than FastPix, by a lot. It is half a cent per gigabyte, flat, worldwide. That same million minutes costs $188 against our $679, so Bunny is 72 percent cheaper and we are not going to pretend otherwise.

The trade-off is in the name of the thing. Bunny's standard network runs 119 points of presence, with 27 in Asia alone. The Volume network runs ten, globally. Fewer points of presence means longer round trips to the edge, and longer round trips show up in one specific place: the time between a viewer pressing play and seeing a frame.

That matters more than it sounds. The Krishnan and Sitaraman study of 6.7 million viewers found that people begin abandoning a video if it does not start within about two seconds, and that each additional second of startup delay raises the abandonment rate by roughly 5.8 percent. A video that froze for one percent of its duration lost five percent of its watch time.

So the Volume network is a real option with a real cost, and the cost is denominated in viewers rather than dollars. Whether that trade is worth it depends entirely on how bad the startup time actually gets for your audience, which brings us to the part of this comparison that has nothing to do with price.

Does Bunny Stream give you the data to run the business?

Bunny Stream's analytics report playback starts, cumulative watch time, views and watch time by country, an engagement heatmap, an engagement score, bandwidth and requests served. That is a reasonable set of audience metrics and plenty of platforms ship less.

What is not in there: startup time, rebuffering ratio, rendition switches, playback failures, and any breakdown by device or carrier.

Which means the trade you just made on the Volume network is one you cannot check. If drop-off climbs on a title, you can see that it climbed. You cannot see whether the first eight seconds took four seconds to start on a mid-range Android handset in São Paulo, or whether the episode was simply not very good. Those two problems get fixed by completely different teams, and one of them is not a content problem at all.

You cannot optimise a metric your vendor does not report. FastPix's Video Data reports 56 dimensions of playback, ten of them yours to define, per session, broken out by geography, device and rendition. That is the difference between knowing your retention dropped and knowing why.

The deeper gap is not quality of experience, though. It is that Bunny reports totals rather than events. Views, watch time and an engagement curve are aggregates: they tell you the number moved, not who moved it or what they did next. Every question a video business actually runs on needs a per-view record carrying a viewer identifier, the title and episode, the playhead position, the custom fields you defined, and the sequence of play, pause and seek events behind them.

That distinction decides what you can build. A microdrama episode funnel and paywall dashboard needs viewer identity carried across sequential episodes to compute a hook rate, and needs the paywall state on each view to tell a content problem, which shows up as a gradual decline, apart from a monetisation problem, which shows up as a cliff at the lock. A video engagement workflow that issues a certificate at ninety percent completion with sixty percent coverage, nudges the learners who opened a lesson and never watched it, and pulls clips from the most-replayed moments, runs entirely on playhead position, play counts, backward jumps and viewer identifiers.

None of those can be assembled from view counts and total watch time, at any level of effort, because the individual records were never captured. Aggregate analytics tell you the business moved. Event data tells you what to do about it.

What else is in the box

Once the price question is settled, the rest of the comparison is about what you would otherwise build. The player is the largest single item in that list, so it gets its own section below.

Two of those rows change what a team can do rather than what it pays. In-Video AI turns captioning from a per-video invoice into a property of upload. Clipping Agent turns one episode into three or four vertical clips with captions burned in, because most people watch social video on mute, and five to ten clips a day is 150 to 300 doorways a month back into your app. Neither appears anywhere in a per-gigabyte comparison, and for most consumer video businesses the second one is worth more than the entire infrastructure line.

The last row is worth its own article, so briefly: Bunny's free encoding tier stops at 1080p in H.264. Anything above that, or in AV1, HEVC or VP9, requires their paid Premium Encoding, billed per output codec. FastPix's Standard tier encodes free at every resolution through 4K. We will do the full comparison separately.

What does a free player actually include?

Both products ship a player at no charge, and on a comparison table that looks like a tie. It is worth pulling apart, because "player" is not a feature. It is a program.

Bunny documents a responsive web embed player: clean interface, consistent across browsers, keyboard navigation and screen-reader support, working on desktops, tablets and mobile browsers. That is a good web player, and for a web-only product it is all you need.

What Bunny does not document is the rest of the surface area a video product eventually needs. Native software development kits (SDKs) for iOS, Android, Flutter and smart TVs are not part of what they publish, and neither are casting, offline downloads or playback telemetry. If your roadmap has a mobile app or a TV app on it, that work lands on your team.

FastPix Player covers the following, and each row is a thing somebody would otherwise build and then maintain across every platform you ship on:

The analytics row is the one that connects back to the previous section. Quality of experience data is not something bolted onto the player afterwards through a third-party integration. It is emitted by the player, which is why it arrives without any work from you and why a platform that does not ship its own player tends not to have it.

A free player is only free until the second platform.

Is Bunny Stream the right choice for a cost-conscious startup?

When delivery really is all you need. A concentrated audience in Europe or North America, a catalogue that does not need to look especially sharp, a player you are happy with, and no requirement for per-session playback data. In that shape Bunny is cheaper on bytes, cheaper on storage by roughly three times, and simpler to reason about, and a team that needs exactly that should buy exactly that.

None of this makes a bad show good. No pricing model has ever fixed a retention problem, and infrastructure does not program taste. What it decides is whether growing your audience is something your finance team looks forward to.

The real version

Take your own numbers. Pull your delivered-rendition split and your audience by region out of your player analytics, blend Bunny's four published rates against your actual traffic, and compare it to a flat per-minute rate. The arithmetic is in this article and it takes an afternoon.

If your audience lives in one cheap region and your bitrate is modest, you will find Bunny is cheaper, and you should act on that. If your audience is spread across the world, or you expect it to be, you will find the gap runs the other way and widens every year you succeed.

Upload it. Encode it. Stream it. Clip it. Read the numbers. Modern video is what we do.