Mbps Converter

Why Mbps and MB/s are eight times apart

Megabits per second and megabytes per second are two names for the same pipe, and the gap between them is a factor of eight. One byte is eight bits, so a plan sold as 100 Mbps delivers 12.5 MB/s once your download manager starts counting in bytes. That single division explains most of the "my internet is slow" complaints that turn out to be unit confusion.

Speed is quoted in bits because that is how the physical link is measured: how many binary digits cross the wire each second. Files are measured in bytes because that is how storage works. A browser download, a Steam install, a phone hotspot counter and a NAS copy all report bytes, usually written MB/s or MBps, while the ISP, the router admin page and the FCC broadband map all report bits, written Mbps or Mb/s.

The capital B is the whole difference. Lowercase b is a bit, uppercase B is a byte, and 8 bits make a byte. Mix them up and a 940 Mbps fiber line looks like it is running at 940 megabytes, which would be 7,520 Mbps of throughput and does not exist on consumer gear.

The one step conversion, and the download time formula

The conversion is one division, and it works in both directions:

• MB/s = Mbps / 8
• Mbps = MB/s x 8

That is the entire formula. 8 Mbps is 1 MB/s. 80 Mbps is 10 MB/s. 400 Mbps is 50 MB/s. If you can halve a number three times, you can do it in your head: halve 400 to 200, halve to 100, halve to 50.

Download time comes from the same numbers. A file size in gigabytes becomes megabits by multiplying by 8,000 (1 GB is 1,000 MB and each MB is 8 Mb), then you divide by the plan speed:

seconds = GB x 8,000 / Mbps

For a 1 GB file on a 100 Mbps plan that is 1 x 8,000 / 100 = 80 seconds. On 500 Mbps it is 8,000 / 500 = 16 seconds. On 1 Gbps it is 8 seconds, because 1 Gbps is 1,000 Mbps and 8,000 / 1,000 = 8.

Storage makers count a gigabyte as 1,000 megabytes, which is why a "1 TB" drive shows about 931 GB in Windows. If you use the 1,024 convention instead, 1 GB is 8,192 Mb and the same 100 Mbps download takes 81.9 seconds rather than 80. The difference is under 3% and it does not change any decision you would make about a plan.

Worked examples: 25 Mbps to 1 Gbps

Four plans, one 1 GB file, the same arithmetic each time:

• 25 Mbps: 25 / 8 = 3.125 MB/s, and 8,000 / 25 = 320 seconds, or 5 minutes 20 seconds
• 100 Mbps: 100 / 8 = 12.5 MB/s, and 8,000 / 100 = 80 seconds
• 500 Mbps: 500 / 8 = 62.5 MB/s, and 8,000 / 500 = 16 seconds
• 1,000 Mbps: 1,000 / 8 = 125 MB/s, and 8,000 / 1,000 = 8 seconds

The jump from 25 to 100 Mbps cuts the wait from 320 seconds to 80, a 75% reduction for four times the speed. The jump from 500 Mbps to 1 Gbps cuts 16 seconds to 8, a 50% reduction for twice the speed. Each doubling of bandwidth halves the time, which sounds obvious until you notice that the last doubling saves 8 seconds and the first one saved 240.

That is the argument for matching the plan to the job. A household that streams, browses and backs up photos to the cloud will not feel the difference between 500 Mbps and 1 Gbps on a single download, because the bottleneck moves to the server on the other end or the Wi-Fi link in the room. A household that moves 200 GB of video files a week will feel it immediately: 200 GB at 12.5 MB/s is 4 hours 26 minutes, and at 125 MB/s it is 26 minutes 40 seconds.

Small files do not scale the same way. Ten thousand 40 KB images take far longer than their total size suggests, because each one pays a round trip of latency. At 30 ms per request, 10,000 requests is 300 seconds of waiting no matter how wide the pipe is.

Kbps, Mbps and Gbps: the steps of 1,000

The ladder runs in steps of 1,000, not 1,024, because network speeds use decimal prefixes. That trips people up when they move between units and storage conventions in the same breath.

So 1 Gbps is 1,000 Mbps and 1,000,000 Kbps. A 300 Kbps audio stream is 0.3 Mbps. A 5 Mbps video call is 5,000 Kbps. When a router admin page reports 867 Mbps on a Wi-Fi 5 link and a laptop shows 54 MB/s in a file copy, those two numbers are consistent: 867 / 8 = 108 MB/s of theoretical link rate, and about half of that survives a real transfer.

The FCC raised its fixed broadband benchmark to 100 Mbps down and 20 Mbps up in March 2024, four times the 25/3 Mbps standard it had used since 2015. That number is a definition of "broadband", not a recommendation for your house, but it is a useful floor: anything under 100 Mbps down is below what the agency counts as advanced capability in 2026.

Why the measured speed never matches the label

A 100 Mbps plan rarely tests at exactly 100. Some of the gap is real loss, some is measurement, and some is a hard ceiling in your own equipment.

Protocol overhead. Every packet carries headers, and TCP acknowledges what it receives. The payload share of a saturated link usually lands between 94% and 97% of the line rate, so 100 Mbps of link speed shows up as roughly 95 Mbps in a speed test and about 11.9 MB/s in a download.

The 100 Mbps port problem. A router or laptop with a Fast Ethernet port negotiates at 100 Mbps and cannot exceed it. Take off overhead and the ceiling is about 94 Mbps. Anyone on a 500 Mbps or 1 Gbps plan who tests at 94 Mbps has found a Fast Ethernet port in the chain, usually an old switch, a cheap USB adapter or a router between the modem and the PC.

Wi-Fi. Radio is half duplex and shares airtime, so a link that negotiates at 867 Mbps delivers somewhere near 400 to 500 Mbps of real throughput in the same room, and less through walls. A 1 Gbps plan tested on Wi-Fi at 300 Mbps is not a broken plan, it is a Wi-Fi limit.

The other end. A download from a single server is capped by that server's upload and by peering. Speed tests use nearby servers for a reason: the number they report is your link, not the internet's willingness to feed one connection.

Provisioning. Many ISPs set the plan slightly above the advertised rate so that speed tests land on the number, which is why a "100 Mbps" plan can show 105.

Those ranges describe what people see, not a guarantee. The useful test is a wired connection straight into the router with nothing else running, which removes three of the five rows above in one step.

Matching a plan to what the house does

Once the units are straight, plan size becomes a question about simultaneous use rather than about one download.

Streaming bitrates are the honest way to size this. An HD stream runs about 5 Mbps, a 4K stream about 15 to 25 Mbps, and a 1080p video call about 2 to 4 Mbps up and down. Four 4K streams at 20 Mbps each is 80 Mbps, which fits a 100 Mbps plan with room to spare on paper and feels tight in practice because of overhead and Wi-Fi loss.

Upload matters more than the download number for anyone working from home. The FCC benchmark pairs 100 Mbps down with 20 Mbps up, and a 20 Mbps upload is 2.5 MB/s, which is fine for a video call and slow for sending a 2 GB video file: 2,000 x 8 / 20 = 800 seconds, or 13 minutes 20 seconds.

Testing your own connection without fooling yourself

Three habits make the numbers trustworthy.

Test wired, once. Plug a laptop into the router with a known good cable, close everything else, and run a test against a server in your city. That number is your plan. Every other test is your plan minus whatever the path costs.

Watch the units in the app. A game launcher showing "12 MB/s" on a 100 Mbps plan is running at full speed, and the same launcher on a 1 Gbps plan showing "12 MB/s" is using a tenth of it. Before you call support, divide the plan by 8 and compare like with like.

Check the link rate, not just the speed test. The router or laptop will report the negotiated Wi-Fi rate, which is the ceiling before overhead. A link rate of 300 Mbps cannot deliver 300 Mbps of file transfer, and knowing that stops a lot of pointless troubleshooting.

If a wired test comes in under 80% of the plan twice in a row, with a modern router and a Cat 5e or better cable, the problem is upstream of your house and worth a call. If the wired test is fine and the Wi-Fi is not, the fix is a mesh node, a different channel or a cable, not a bigger plan.

Speed definitions and benchmarks change, and the FCC reviews its broadband standard periodically. Confirm current figures with the FCC before relying on them for a purchase or a complaint.