AssemblyAI Review: Speech-to-Text API Pricing and Models
A transcription API for developers, priced per hour of audio.
AssemblyAI is a speech-to-text API for developers who build transcription into their own apps. It is not a meeting app. Transcription costs from $0.15 an hour of audio, and new accounts get $50 of free credits.
Checked November 2026 on assemblyai.com.
AssemblyAI models and prices
| Model | Price per hour of audio | Use |
|---|---|---|
| Universal-3.5 Pro | Not published | Recorded audio, 18 languages |
| Universal-2 | $0.15 | Recorded audio, 99 languages |
| Universal-Streaming | $0.15 | Live audio, English or multilingual |
| Universal-3.6 Pro Realtime | $0.45 | Live audio for voice agents |
Strengths and limits
Strengths
- • Low price per hour, billed as you go
- • 99 languages on Universal-2
- • Speaker labels from $0.02 an hour
Limits
- • You need to write code to use it
- • No bot that joins meetings
- • Streaming is billed for open time, not audio sent
Is AssemblyAI free?
AssemblyAI gives $50 of free credits at sign-up, with no credit card. At $0.15 an hour of audio, that covers about 330 hours on Universal-2. After that you pay as you go.
How much does AssemblyAI cost?
Pre-recorded audio costs $0.21 an hour on Universal-3.5 Pro and $0.15 on Universal-2. Live streaming costs $0.15 an hour, or $0.45 on Universal-3.6 Pro Realtime, and is billed for the time the connection stays open. Add-ons such as speaker labels cost extra, from $0.02 an hour.
What is AssemblyAI used for?
Developers use it to add transcripts to their own products: meeting note takers, call analytics, captions and voice agents. The API returns text with speaker labels, and add-ons detect entities, redact personal data or translate.
Is AssemblyAI a meeting assistant?
No. It does not join Zoom, Google Meet or Teams calls, and it has no app for notes. To get meeting notes without writing code, use a meeting tool such as Otter or Fireflies.