File Size Calculator
Convert file sizes between bytes, KB, MB, GB, TB. Calculate transfer times at different speeds. Free online file size converter
Asking "how big is 5 GB" without saying which "GB" leads to a 7% error — and 7% on a 10 TB drive is 700 GB, which is the difference between "it fits" and "it does not". The 1024 vs 1000 split is the source of every "I bought a 1 TB drive and Windows shows 931 GB" complaint. This converter handles both definitions, shows the canonical units (KiB / MiB / GiB for 1024-based, KB / MB / GB for 1000-based), and computes transfer times at common connection speeds so you know whether a download will finish during lunch or overnight.
KB vs KiB and why this matters
- Decimal (SI) units — kB = 1000 B, MB = 1,000,000 B, GB = 10^9 B, TB = 10^12 B. Used by hard drive manufacturers (so a "1 TB drive" has 1,000,000,000,000 bytes), network speed quotes (Mbps, Gbps), and most consumer marketing.
- Binary (IEC) units — KiB = 1024 B, MiB = 1024² B, GiB = 1024³ B, TiB = 1024⁴ B. Used by RAM (RAM addresses are powers of 2), Linux/macOS file managers, and most "how much memory does this process use" tools.
- Microsoft "GB" — Windows displays GiB but calls it "GB". This is why a "1 TB" drive shows "931 GB" in Windows: the drive has 10^12 bytes = 931.32 GiB, and Windows shows the GiB number labeled as GB.
- Mixed mode in network speed — "1 Gbps internet" is 1 gigabit per second (10^9 bits/s) = 125 MB/s = 119 MiB/s. The factor-of-8 between bits and bytes catches everyone the first time.
Working example: download time at common speeds
Input
File: 4.7 GiB (a typical DVD image)
Output
File size in different units: 4.7 GiB = 5.05 GB (decimal) = 5,046,586,572 bytes Transfer time: 56k modem (56 kbps) : 8.4 days — historical 3G (1 Mbps real) : 11.2 hours Cable (50 Mbps) : 13.5 minutes Fiber (1 Gbps) : 40 seconds (theoretical) / 60s real-world Gigabit LAN (1 Gbps) : 40 seconds — same theoretical max as fiber USB 3.0 (5 Gbps) : 8 seconds (theoretical) / 30s real-world USB 4 / Thunderbolt 4 : 1.5 seconds (theoretical) / 5s real-world Local NVMe SSD copy : 1 second Real-world ≈ theoretical × 0.5-0.7 (TCP overhead, disk write speed, encryption, fragmentation).
The gap between "Gigabit fiber should take 40 seconds" and "actually takes 2 minutes" is overhead. TCP slow-start, server-side throttling, browser HTTP/2 multiplexing, your CPU's decryption speed, and disk I/O all contribute. For sizing-the-coffee-break decisions, multiply theoretical by 1.5-2.
Network speed gotchas
- Bits vs bytes — network speeds are quoted in bits per second (Mbps, Gbps). File sizes in bytes. Divide bps by 8 to get bytes per second. 1 Gbps = 125 MB/s.
- Advertised vs sustained — ISP "up to 1 Gbps" means peak; sustained over hours might be 70-80% of that, less during congestion windows. Speed-test apps measure short bursts.
- Wi-Fi vs Ethernet — Wi-Fi advertised speeds (e.g., "Wi-Fi 6 9.6 Gbps") are total channel capacity shared by all devices; per-device real throughput is much lower. For sustained large transfers, Ethernet is more predictable.
- Symmetric vs asymmetric — most home connections are asymmetric: 1 Gbps down, 50 Mbps up. Upload speed is what matters when sending files (uploading to cloud, backup, video calls).
- Server-side caps — even on a 1 Gbps connection, downloading from a server limited to 10 MB/s gets you 10 MB/s. The internet between you and the server is the bottleneck if either side is slow.
When to reach for this tool
- You are sizing a backup or migration and want to know how long the transfer will realistically take.
- You are sizing cloud storage or buying a hard drive and want to compare "advertised" capacity to "actually usable" capacity after filesystem and OS reporting.
- You are debugging "the file is 1 GB but my disk shows it taking 1.07 GB" — usually filesystem allocation block sizes for small-file overhead, or KB vs KiB confusion.
- You are estimating bandwidth costs for a SaaS / CDN — converting "10 TB egress per month" into "average MB/s" or "GB/day" for capacity planning.
What this tool will not do
- It will not measure your actual disk usage. For that, use du -h (Linux/macOS) or a disk usage analyzer (WinDirStat, DaisyDisk).
- It will not measure your actual network speed. Run a speed test (fast.com, speedtest.net) for that. Speed estimates here are theoretical based on the link rate you provide.
- It will not account for protocol overhead in real connections. HTTP/2 vs HTTP/3, TLS handshake, TCP window scaling — all affect real-world transfer time. The estimates here are good to ~30% accuracy; for tight planning, test with the actual transfer.
Frequently asked questions
Why does my 1 TB hard drive show 931 GB?
Drive manufacturer used 1 TB = 10^12 bytes. Windows displays that as 931 GiB (10^12 / 1024^3) but labels the unit "GB". The bytes are not missing; the labels are inconsistent. macOS since 2010 uses decimal units to match drive labels — your "1 TB" drive shows "1 TB" on macOS.
How big is "1 KB of text"?
~1024 characters in ASCII or basic UTF-8 (one byte per character for English). For UTF-8 with mixed Latin / CJK / emoji, anywhere from ~500 to 1000 characters per KiB. A page of double-spaced 12pt text is roughly 2 KB.
Is 1 Mbps faster than 1 MB/s?
No. 1 MB/s = 8 Mbps. People confuse these constantly. "10 Mbps internet" downloads at ~1.25 MB/s, which means a 100 MB file takes ~80 seconds. The factor of 8 makes "I have fast internet" feel slower than the marketed speed.
What is a "petabyte" — does anyone use that in real life?
PB = 1000 TB (decimal) or PiB = 1024 TiB (binary). Used in big-data context: a major search engine indexes ~150+ PB; Netflix's entire content library is ~10s of PB; the Library of Congress's text content is ~20 PB. Most individuals will never personally own PB-scale storage; petabytes are enterprise/datacenter units.
How do I tell file size before downloading?
The HTTP Content-Length header (sent in the response headers, visible in browser devtools). For large downloads, the server may use chunked transfer encoding and not send Content-Length — in that case, size is unknown until complete.
Why does copying a small file take "0 seconds" but feels longer?
Filesystem overhead. Creating a file requires updating directory entries, file allocation tables, journaling logs — fixed cost per file regardless of content. A 1 KB file might take 5 ms to create even though writing the bytes takes microseconds. For many tiny files, the overhead dominates: copying 10,000 1-KB files is much slower than copying one 10-MB file.
Related tools
Free online unit converter. Convert length (meters, feet, inches), weight (kg, lbs), temperature (Celsius, Fahrenheit), volume, area, speed and more units instantly
Compress and resize images online without losing quality. Reduce JPEG, PNG, WebP file size. Free image optimization tool with adjustable quality
Generate MD5, SHA-1, SHA-256, SHA-512 hashes from text or files. Verify file checksums. Free online hash calculator and checksum generator
Calculate MD5, SHA-1, SHA-256, SHA-512 checksums for files. Verify file integrity and compare hashes. Free online file hash calculator
Calculate and convert image/video aspect ratios. Find dimensions for 16:9, 4:3, 21:9 and custom ratios. Free online aspect ratio calculator
Convert images between PNG, JPEG, and WebP formats. Batch conversion with quality control. Free online image format converter
Last updated · E-Utils editorial team