Convert.FAST

How it works

Why are conversions so fast?

We run our own servers, control the entire stack, and obsess about every millisecond.

We own the full stack

We run dedicated servers in the European Union and control every layer—database, application, storage, network ingress and egress. No cloud provider middlemen. No shared resources. This lets us optimise the entire pipeline from upload to download.

When you control everything, you can tune everything. We eliminate bottlenecks most services cannot touch because they sit in someone else's infrastructure.

We use the best tools for the job

We use a carefully selected mix of open-source and commercial processing libraries—multi-threaded, streaming-based, and optimised for each format. Our image pipeline is 10-20× faster than typical web converters for common operations like HEIC, JPG, PNG, WebP, and AVIF.

Example: A 12MP iPhone HEIC photo (4032×3024, ~3 MB) converts to JPG in ~0.3 seconds on our infrastructure.

Trade-off: AVIF encoding takes longer (4-8 seconds) because AV1 is computationally expensive. We think the quality and file size gains are worth it, but we are transparent about the time cost.

We built a fair scheduler

Our custom job scheduler ensures everyone gets fast processing, even during peak load. Paid users get priority (Business, Pro, then Free, then Anonymous), but no single user can monopolise workers. Jobs are distributed fairly within each tier using round-robin rotation, not first-come-first-served.

If a Pro user uploads 100 files, only 6 process concurrently. This leaves capacity for other users instead of blocking the queue.

We sweat the details

Speed is not a feature—it is the architecture. Files stream directly from your browser to our servers without intermediate buffering. Jobs queue in memory with sub-millisecond pickup times. We enforce a 100-megapixel limit to prevent resource exhaustion while supporting everything from phone photos to gigapixel panoramas.

For most EU users, network round-trip latency is 10-30ms. We cannot make the speed of light faster, but we can eliminate every other delay we control.


Real-World Performance

Typical end-to-end times (upload → conversion → download ready):

  • Small JPG (1-2 MB, 3000×2000) — 0.5-1.0 seconds total
  • iPhone HEIC (3-5 MB, 4032×3024) — 0.3-0.6 seconds total
  • Large PNG (10-20 MB, 6000×4000) — 1.5-3.0 seconds total
  • AVIF encoding (any size) — 4-8 seconds (AV1 is computationally expensive)
  • 1-hour M4B audiobook → OGG — ~15 seconds (~250ms per minute of content)
  • 3-hour audiobook → MP3 — under 1 minute

These are median (p50) times. p95 times are typically 2-3× higher during peak load.