Skip to content
Blogiantic

AssemblyAI

AI models to transcribe and understand speech

FreemiumChecked

AssemblyAI is a speech-to-text service for developers that transcribes audio and video through APIs, with speaker labels, timestamps and support for over 99 languages. A free tier needs no credit card and includes up to 185 hours of pre-recorded transcription. Pay-as-you-go rates are listed from $0.15 per hour on its pricing page. It suits teams building transcription into their own products.

Visit assemblyai.com
AssemblyAI homepage, captured Nov 10, 2026
assemblyai.com homepage, captured Nov 10, 2026.

What AssemblyAI does

AssemblyAI sells speech recognition as a set of APIs that developers build into their own products. The Pre-recorded Speech-to-Text API takes audio and video files and returns transcripts with speaker labels and formatting, while separate APIs handle streaming audio in real time, short clips returned in the same request, and dictation. The pricing page lists language detection, formatting, filler words, keyterms prompting, custom spelling and word-level timestamps, and the company states support for over 99 languages across its models.

Speaker Diarization tells speakers apart, and Speaker Identification can swap generic labels for names or roles. Beyond transcripts, a Speech Understanding API adds summaries, sentiment and entities, Guardrails detect and redact PII and unsafe content, and a Text-to-Speech API is also offered. A medical mode adds clinical vocabulary, a no-code playground lets you test models, and models can be self-hosted in your own VPC or on-prem.

Billing is monthly on actual usage, with no minimum commitments, upfront fees or contracts on pay-as-you-go. The two pre-recorded models are listed at $0.21 and $0.15 per hour, and multichannel audio is billed per channel. Custom pricing covers higher rate limits and volume discounts.

Features

  • Pre-recorded Speech-to-Text API for audio and video files
  • Realtime Speech-to-Text API for streaming audio
  • Sync Speech-to-Text API and Dictation API
  • Speaker diarization and speaker identification
  • Language detection, formatting, custom spelling and word-level timestamps
  • Speech Understanding API: summaries, sentiment and entities
  • Guardrails to detect and redact PII and unsafe content
  • Text-to-Speech API
  • Medical mode for clinical vocabulary
  • No-code playground for testing models
  • Self-hosted deployment in your own VPC or on-prem

AssemblyAI pricing

Checked
AssemblyAI plans and prices as listed on its pricing page
PlanPriceWhat's included (as stated)
Universal-3.5 Pro$0.21 usagebilling term not confirmed
  • $0.21 /hr, pay as you go
  • Pre-recorded speech-to-text
  • 18 languages, with native code switching
Universal-2$0.15 usagebilling term not confirmed
  • $0.15 /hr, pay as you go
  • Pre-recorded speech-to-text
  • Supports 99 languages
CustomCustom
  • Custom rate limits and enhanced concurrency
  • Volume-based pricing

Copied from AssemblyAI's pricing page on Nov 10, 2026. Prices change, so confirm before you buy.

Strengths and limits

Each point comes from the features and plan details on this page. No point is an opinion.

Strengths

  • Pay-as-you-go transcription from $0.15 per hour, with no minimum commitments or contracts.
  • Free tier needs no credit card and includes up to 185 hours of pre-recorded transcription.
  • Speaker Diarization labels each speaker, and Speaker Identification can swap the generic labels for names or roles.
  • Models can be self-hosted in your own VPC or on-prem.

Limits

  • Transcription is used through APIs, so it needs development work to put into use.
  • Multichannel audio is billed per channel, so each channel is charged at the model's hourly rate, such as $0.15 /hr on Universal-2.
  • The Universal-3.5 Pro model covers 18 languages, far fewer than Universal-2.

Who AssemblyAI suits, by the numbers

Best for:
trying it at no cost (free plan: Up to 185 hours of pre-recorded transcription); building it into your own software through its API.

Signals used: free plan, platforms, as verified on Nov 10, 2026. How signals work.

What people use AssemblyAI for

AssemblyAI FAQ

Is AssemblyAI free?

There is a free tier that needs no credit card. It includes up to 185 hours of pre-recorded transcription and up to 333 hours of streaming transcription.

How much does AssemblyAI cost?

Pre-recorded transcription is listed at $0.21 per hour for Universal-3.5 Pro and $0.15 per hour for Universal-2, billed monthly on actual usage. Custom pricing is available for high volumes.

Where this listing comes from

Facts were read from these pages on Nov 10, 2026. The description was written from those facts. See the methodology.

Something out of date? Report a correction. Corrections are free.

Similar tools

AssemblyAI alternatives