Is AssemblyAI free?

AssemblyAI is free to start and paid to scale. Its free tier includes up to 185 hours of pre-recorded transcription and up to 333 hours of streaming transcription ¹. You can create an account and start transcribing immediately, with no credit card required ². When the allowance runs out, you pay by the hour of audio. This page reflects the company's pricing page as read on 27 October 2026, and the company can change the tier.

The free tier is a testing allowance, and it has limits

The allowance is large for a prototype. At the Universal-3.5 Pro list price of 0.21 dollars per hour ³, the 185 pre-recorded hours would otherwise cost about 38.85 dollars ⁴. That is a lot of test audio, but it is a one-time allowance for testing, not a monthly plan for running a product.

Two practical limits to know:

• Streaming connections. On the free plan you can open up to 5 new streaming connections per minute ⁵. That is fine for development and too low for an app with many simultaneous users.
• Pre-recorded versus streaming. The allowance is stated separately for files (185 hours) and live audio (333 hours), so plan your test for the one you will use.

The pricing page is the authority on any other limit, such as how long an account keeps the free allowance. Read it before you plan around the free tier.

What you pay after the free tier

There is no minimum commitment, upfront fee or contract on pay as you go, and billing is monthly for the previous month's usage ⁶. The batch rates are 0.15 dollars per hour for Universal-2 ⁷ and 0.21 dollars per hour for Universal-3.5 Pro. Add-ons such as speaker labels cost extra, and a stereo file counts as two hours. Our question on how much AssemblyAI costs has the full table.

Free does not mean private

The free tier does not change the question of what happens to your audio. For pre-recorded audio, the company advises a short time-to-live, or deleting transcripts through the API, because it does not offer zero retention for that product ⁸. If your audio is sensitive, read the retention terms in the guide to AssemblyAI and our question on whether speech-to-text APIs store audio before you upload it.

Free ways to use AssemblyAI without writing code

An API is for developers, so the free tier alone will not give you an app. A few routes exist:

• Third-party apps on your own key. Blurt, an open-source dictation app, is free under the MIT license ⁹. It runs on AssemblyAI's Dictation API, and users bring their own key and pay for what they transcribe, around 0.62 dollars per hour of audio ¹⁰. The app is free, but the transcription is not.
• The API directly. If you can run a short script, the free tier is enough to transcribe a few days of recordings.

When another free option fits better

It depends on what you need.

• A one-off transcript with no account. A browser tool such as audio to text transcribes a file for free.
• Free speech-to-text APIs. Compare the allowances in our guide to the free speech-to-text APIs.
• No per-hour fee at all. The open Whisper model is free to run on your own computer, as the question on whether Whisper is free explains, at the cost of your hardware and setup time.
• Accuracy first. See AssemblyAI against Whisper for how the two compare.

What to do

Sign up, run 30 to 60 minutes of your own audio through both batch models, and note how many free hours the test used. If the accuracy is good enough, estimate your monthly hours at the paid rates before the allowance ends. The speech recognition overview lists alternatives, and for on-device dictation that needs no API, Yap, which we make, can be tried from the download page.