AssemblyAI vs Whisper API on price

AssemblyAI or Whisper API: quick answers

Is AssemblyAI cheaper than Whisper API?

For recorded audio, AssemblyAI is cheaper at every volume from 1 minute to 10,000 hours a month. At 1,000 hours a month that's $210 a month apart, $2,520 a year ($150 on AssemblyAI's Universal-2 against $360 on Whisper API's Whisper (whisper-1)). For live audio, AssemblyAI is cheaper at every volume from 1 minute to 10,000 hours a month. At 1,000 hours a month that's $870 a month apart, $10,440 a year ($150 on AssemblyAI's Universal-Streaming against $1,020 on Whisper API's gpt-realtime-whisper). Regular pay-as-you-go prices, each side's cheapest model.

How much more does live transcription cost than recorded on AssemblyAI and Whisper API?

AssemblyAI charges the same for both, $0.15 an hour (Universal-2 and Universal-Streaming); on Whisper API, live audio costs 183% more than recorded, $1.02 an hour on gpt-realtime-whisper against $0.36 on Whisper (whisper-1). Each side's cheapest model for each kind of audio at 100 hours a month, at its list price. AssemblyAI bills live audio by how long the connection stays open, idle time included.

Do AssemblyAI and Whisper API have a free tier or free credit?

AssemblyAI: AssemblyAI says “You can create an account and start transcribing immediately with no credit card required. The free tier includes up to 185 hours of pre-recorded transcription and up to 333 hours of streaming transcription.” and “$50 in free credits, no credit card required.” Whisper API: No free allowance or credit is listed on OpenAI's pricing page.

Do AssemblyAI and Whisper API bill per second?

AssemblyAI: AssemblyAI says “Pre-recorded is billed per hour of audio submitted.” and “Streaming is billed per session duration — the time the WebSocket connection is open, not the duration of audio sent. Idle connection time counts.” Whisper API: OpenAI's pricing page doesn't say how partial minutes are counted.

Does stereo audio cost double on AssemblyAI or Whisper API?

AssemblyAI: “Multichannel audio is billed per channel. For example, a 1-hour stereo (2-channel) file is billed as 2 hours of transcription.” Whisper API doesn't say how it bills more than one channel on the pages we read. So 100 hours of two-channel recorded calls a month cost $30 on AssemblyAI (Universal-2, 200 billed hours).

What do speaker labels add on AssemblyAI and Whisper API?

For recorded audio: AssemblyAI charges $0.02 an hour extra; Whisper API offers them through the separate gpt-4o-transcribe-diarize model, at the same estimated $0.006 a minute as gpt-4o-transcribe. For live audio: AssemblyAI charges $0.12 an hour extra; Whisper API's pricing page doesn't list them.

Do AssemblyAI or Whisper API give volume discounts?

AssemblyAI: Volume discounts are custom, through sales. Whisper API: no volume discount is listed on its pricing page.