Why BYOK?
FlashTranslate.Net doesn’t run its own translation servers, and it doesn’t mark up your usage. You connect your own API key from an LLM provider (OpenAI, Google Gemini, OpenRouter, or any OpenAI-compatible endpoint), and every translation request goes straight from your browser to that provider — nothing in between.
That gives you two things most translation extensions can’t: real speed, and real cost transparency.
Fast: no middleman, no queue
Section titled “Fast: no middleman, no queue”Because requests go directly to the provider you chose, there’s no relay server adding latency or getting rate-limited by shared traffic. You’re only as fast as your chosen model’s own API — and the fast/cheap models recommended below typically start streaming a translation back in well under a second.
Cheap: here’s what it actually costs
Section titled “Cheap: here’s what it actually costs”“Pay per token” sounds abstract until you see it against something concrete. Here’s the estimated cost of a few realistic use cases, calculated from the token prices FlashTranslate.Net ships with today (see the table below) — prices move with each provider’s own pricing pages, so treat these as illustrative, not a quote.
| Scenario | Google Gemini 2.5 Flash Lite | OpenAI GPT-5.4 Nano |
|---|---|---|
| a 10-minute YouTube video’s subtitles (~1,500 words) | ~$0.001 (almost free) | ~$0.003 |
| a 1-hour podcast/meeting transcript (~9,000 words) | ~$0.007 | ~$0.02 |
| a full academic paper (~10,000 words) | ~$0.008 | ~$0.02 |
| Alice in Wonderland in full (~30,000 words) | ~$0.02 | ~$0.07 |
OpenRouter pricing tracks whichever underlying model you pick through it — usually close to that model’s direct price. A custom provider’s cost depends entirely on the rates it publishes.
Get your own key in 3 minutes
Section titled “Get your own key in 3 minutes”If those numbers look worth it, the next step is getting an API key from a provider — it’s a short, one-time setup. See Get an API key in 3 minutes for a step-by-step walkthrough.