Question 1

Is Inferentium really OpenAI-compatible?

Accepted Answer

Yes. Point your existing client at https://api.inferentium.ai/v1 and call audio.transcriptions.create as usual. Only the base_url and the API key change — no other code changes.

Question 2

How much does transcription cost?

Accepted Answer

It depends on which product you use, because they are priced differently: pre-recorded audio is US$0.05 per hour, Sync (transcript in the same response, up to 2 minutes) and Realtime are US$0.12 per hour. All of them are billed per second of audio via prepaid credits, with no subscription and no minimums.

Question 3

Which languages are supported?

Accepted Answer

25 languages with automatic language detection, including English, Spanish, Portuguese, French, German, Italian, Dutch and Japanese.

Question 4

How accurate is it in English?

Accepted Answer

We publish it per language rather than as one global claim, because accuracy is a property of the language. On English benchmark audio (FLEURS en-US) short clips come back at 2.1% word error rate; the full table, including Portuguese and the longer duration classes, is on the model spec sheet — along with how each number was measured.

Question 5

How fast is it?

Accepted Answer

It depends on length, so here are both ends. A 90-minute file transcribes in about 25 seconds — roughly 230× realtime — because long audio amortises the fixed cost every request pays. A short clip is dominated by that fixed cost instead: expect around a second end to end, of which the model itself is typically 150-300 ms depending on length, and the rest is network.

Question 6

Can I transcribe very long files?

Accepted Answer

Yes. Submit an asynchronous job and we call your webhook when it is ready — built for recordings that run for hours.

Question 7

Do you train on my audio?

Accepted Answer

No. Your audio is used only to produce your transcript and is never used to train any model.

New accounts are by invitation