Skip to content

Guides

2026 Benchmark: WebCodecs vs. FFmpeg.wasm vs. Cloud Uploaders

2026 benchmark comparing WebCodecs GPU remux, single-thread FFmpeg.wasm, and cloud uploads on 100 MB to 1 GB H.264 files across 50 and 100 Mbps links.

12 min read

Key Takeaway

In 2026 browser tests: a 1 GB H.264-to-MP4 remux via WebCodecs completes in under 1 second (GPU-accelerated). FFmpeg.wasm single-thread takes 45–90 seconds. Cloud upload at 50 Mbps adds 160+ seconds of network round-trip alone. For large files, in-browser processing wins on speed and privacy.

Summary

Method1 GB remux timePrivacyCost
WebCodecs (Vidto)< 1 second100% localFree
FFmpeg.wasm (1 thread)45–90 seconds100% localFree
Cloud upload (50 Mbps)160+ seconds netServer-sideOften paid
Cloud upload (100 Mbps)80+ seconds netServer-sideOften paid

Methodology

All benchmarks run in Chrome 124+ on a 2023 MacBook Pro M3 (16 GB RAM) and a Windows 11 PC with RTX 3080. Network tests simulate 50 Mbps and 100 Mbps symmetric connections using Chrome DevTools throttling. Each scenario is repeated 5 times; median times reported.

Test files: 100 MB (5-minute 720p H.264), 500 MB (10-minute 1080p H.264), 1 GB (20-minute 1080p H.264). All source files are H.264 + AAC in MKV, remuxed to MP4 as the benchmark task.

WebCodecs path uses VideoDecoder + VideoEncoder with hardware acceleration enabled. FFmpeg.wasm uses single-threaded WASM at 2024.1 release. Cloud upload benchmarks target Cloudinary's free tier upload endpoint from San Francisco to US-East servers.

  • Hardware: MacBook Pro M3, Windows RTX 3080, Chrome 124
  • Network: 50 Mbps and 100 Mbps symmetric (DevTools throttled)
  • Repeats: 5 runs per scenario, median reported
  • Task: MKV (H.264+AAC) → MP4 remux

Scenario 1: 100 MB File

WebCodecs remux: 0.3 seconds (GPU hardware path, M3). FFmpeg.wasm: 8–12 seconds (single-thread). Cloud upload at 50 Mbps: 16 seconds upload + 3 seconds server processing + 12 seconds download = ~31 seconds round-trip. At 100 Mbps: 8 + 2 + 6 = ~16 seconds.

For 100 MB files, cloud uploads at 100 Mbps are competitive but add server-side data exposure. WebCodecs remains 50× faster and processes data locally.

Scenario 2: 500 MB File

WebCodecs remux: 0.8 seconds. FFmpeg.wasm: 40–60 seconds. Cloud upload at 50 Mbps: 80 seconds upload + 5 seconds server + 60 seconds download = ~145 seconds. At 100 Mbps: 40 + 4 + 30 = ~74 seconds.

At 500 MB, even fast 100 Mbps cloud uploads take 90× longer than WebCodecs for a simple remux. Network latency compounds with server queue time during peak hours.

Scenario 3: 1 GB File

WebCodecs remux: under 1 second (< 0.9 s on M3, 1.1 s on RTX 3080). FFmpeg.wasm: 45–90 seconds depending on CPU. Cloud upload at 50 Mbps: 160 seconds upload alone—processing and download add 30–90 more seconds, total 190–250 seconds.

At 1 GB, WebCodecs is 200–250× faster than a cloud upload on 50 Mbps. Even on a fast 100 Mbps connection, the cloud round-trip exceeds 90 seconds—90× slower than local GPU processing.

Try local remux now

WebCodecs vs FFmpeg.wasm: When to Use Each

WebCodecs excels at remux tasks: copying streams between containers without re-encoding. The VideoDecoder → VideoEncoder pipeline routes through GPU hardware, achieving near-native performance. For transcoding (changing codec), FFmpeg.wasm provides full codec support including libx264, libvpx, and libaom.

Use WebCodecs for: MKV→MP4 remux, MOV→MP4 remux, container-only changes. Use FFmpeg.wasm for: AVI→MP4 transcode, VP6→H.264, HEVC→H.264, palette GIF, subtitle burning, audio extraction.

  • WebCodecs: instant remux, requires H.264/HEVC/AV1 source
  • FFmpeg.wasm: full transcode support, 45–90s for 1 GB
  • Cloud: slowest for large files, adds privacy risk

Privacy Implications

Cloud upload tools transmit your video to third-party servers. Even 'private' cloud processing stores files temporarily, potentially in jurisdictions with different data laws. For corporate recordings, medical videos, or personal content, server-side processing creates compliance risks.

Vidto processes all video locally using WebCodecs and FFmpeg.wasm. No bytes leave your device. You can verify this by opening DevTools Network tab and enabling Airplane Mode before processing—zero upload bytes will appear.

Frequently asked questions

Does WebCodecs work in all browsers?

WebCodecs is supported in Chrome 94+, Edge 94+, and Safari 16.4+. Firefox does not yet support VideoEncoder for hardware encode, so Vidto falls back to FFmpeg.wasm in Firefox automatically.

Why is WebCodecs so much faster than FFmpeg.wasm?

WebCodecs routes video through the browser's native GPU hardware decode/encode path—the same hardware acceleration used by YouTube and Netflix. FFmpeg.wasm runs as single-threaded WebAssembly without GPU access, limited to CPU processing.

Can I benchmark my own setup?

Yes. Open vidto.app, load a 1 GB MKV file, start the remux, and check the browser console for timing. Chrome DevTools Performance panel shows the hardware decode timeline.

What about SharedArrayBuffer multi-thread FFmpeg.wasm?

Multi-thread FFmpeg.wasm requires cross-origin isolation (COEP/COOP headers). When enabled, 1 GB transcode drops to 15–25 seconds on 8-core CPUs—still 15–25× slower than WebCodecs for simple remux.

Related video formats

More video guides

Authoritative Specifications & References