AssemblyAI Pricing 2026: Model Rates and Add-On Cost Calculator
AssemblyAI pre-recorded transcription costs $0.15–$0.21/hr. See exact speaker-label and redaction totals, streaming billing, and API vs web-tool costs.

AssemblyAI charges $0.15 per audio hour for Universal-2 and $0.21/hour for Universal-3.5 Pro pre-recorded transcription. Standard speaker diarization adds $0.02/hour, making the respective totals $0.17 and $0.23/hour. The right estimate starts with a named model and adds only the features your application enables.
Prices checked on October 5, 2026, in USD against AssemblyAI's official pricing page. Examples below use one-channel pre-recorded audio at published pay-as-you-go rates, excluding free credits, tax, negotiated discounts, and token-based LLM Gateway calls. Pre-recorded STT has the same published rate in the US and EU; the pricing page's 10% in-region note is in the LLM Gateway section and must not be applied to every transcription model.
Quick answer: AssemblyAI transcription and add-on costs
- Pre-recorded models: Universal-2 at $0.15/hour; Universal-3.5 Pro at $0.21/hour.
- With standard diarization: $0.17 or $0.23/hour, respectively.
- Worked feature stack: Base + diarization ($0.02) + entity detection ($0.08) + PII text redaction ($0.08) = $0.33/hour for Universal-2 or $0.39/hour for Universal-3.5 Pro.
- Target Audience: Developers building custom AI apps through API integrations.
- Web alternative: TranscribeBee has a $2.00/hour usage rate with speaker labels and exports; the API has a lower unit price, while the web tool avoids integration work.
The base rates
| Model | Per hour | Use case |
|---|---|---|
| Universal-2 | $0.15 | Pre-recorded audio |
| Universal-3.5 Pro | $0.21 | Pre-recorded audio |
| Universal-Streaming | $0.15 | Streaming session duration |
| Universal-3.6 Pro Realtime | $0.45 | Streaming session duration |
Streaming is billed for the time the connection stays open, including idle time. Its diarization add-on is $0.12/hour, so do not reuse the $0.02 pre-recorded add-on in a real-time estimate. A streaming hour and an hour-long uploaded file can have different billing bases.
Calculating feature totals
Selected add-ons for the pre-recorded examples:
| Add-on | Per hour |
|---|---|
| Standard speaker diarization | $0.02 |
| Entity detection | $0.08 |
| PII text redaction | $0.08 |
| PII audio redaction | $0.05 |
For Universal-2, the example is $0.15 + $0.02 + $0.08 + $0.08 = $0.33/hour. Changing only the base model to Universal-3.5 Pro gives $0.21 + $0.02 + $0.08 + $0.08 = $0.39/hour. At 10 audio hours, these stacks cost $3.30 and $3.90. They are examples of chosen features, not a required minimum for every application.
Speaker diarization assigns labels to turns; the separately listed Speaker Identification feature identifies names or roles. Text and audio redaction are also separate features. Pin the precise option in your estimate, and check current model support before adding summaries, medical mode, or prompting.
Who AssemblyAI is actually for
AssemblyAI is suited to applications that need programmable transcripts and selected analysis features. Model selection, enabled features, channels, and streaming connection duration make the bill reproducible when those inputs are recorded.
When comparing it with an upload service, account for:
- The upload, review, and export workflow your users will need beyond an API or testing playground.
- The add-ons you actually enable rather than every available feature.
- Retries, job handling, file management, and integration maintenance.
When simplicity wins
If you want transcripts without building a pipeline, TranscribeBee provides speaker labels, timestamps, and TXT/SRT/DOC/PDF export at a $2 per audio hour usage rate. Test volume scenarios on our transcription cost calculator, then compare the workflow and features you need.
For 10 hours, a $2/hour usage estimate is $20. Compared with the Universal-3.5 Pro example above, the difference is $20 − $3.90 = $16.10. That is a workflow convenience tradeoff, not proof that a web service is always cheaper overall. Existing integrations and your own labor costs can change the decision.
Estimating your real AssemblyAI bill
Before deciding, run your scenario through the AssemblyAI Cost Calculator prompt in our free AI prompts library — it walks through which add-ons your use case needs and computes the stacked per-hour rate against alternatives.

More Posts

Per-minute, subscription, freemium, and enterprise transcription pricing decoded — what each model really costs per hour and which one fits your usage pattern.

Deepgram Nova-3 runs $0.0043/min batch and $0.0077/min streaming. The real decision is which you need — and whether you need a developer API at all.

OpenAI's Whisper API costs $0.36/hour; self-hosting starts around $861/month all-in. Where the break-even really sits and who should choose what.
Newsletter
Join the community
Subscribe to our newsletter for the latest news and updates