AssemblyAI review

AssemblyAI splits its API into pre-recorded and streaming models and prices both by the hour. For file transcription, Universal-2 covers 99 languages at $0.15 per hour, while Universal-3.5 Pro costs $0.21 per hour and focuses on 18 languages with code switching and stronger diarization. For live audio, Universal-Streaming costs the same $0.15 per hour for English, a multilingual version adds Spanish, French, German, Italian and Portuguese at the same price, and Universal-3.6 Pro Realtime costs $0.45 per hour for 32 languages with about 150 ms median latency, context carryover and voice isolation. Streams return partial and final transcripts with endpointing and word timestamps, and AssemblyAI lists unlimited concurrent streams. The catch is the add-on menu: keyterms prompting, diarization, medical mode and speaker focus each carry a per-hour fee on some models. Python and JavaScript SDKs keep the WebSocket setup short.

Who it is for

Developers building voice agents, call analytics or AI notetakers who want low streaming prices and published latency.

Price

Pay as you go from $0.15 per hour (about $0.0025 per minute) for Universal-2 batch and Universal-Streaming; Universal-3.5 Pro $0.21 per hour; free to start.

Exceptional accuracy at a lower price. $0.15 /hr
Quoted from the vendor pricing page on November 7, 2026.

Key features

• Universal-2 batch model in 99 languages and Universal-3.5 Pro in 18 languages with code switching
• Universal-3.6 Pro Realtime for voice agents in 32 languages with about 150 ms p50 latency
• Partial and final transcripts, intelligent endpointing and word-level timestamps
• Keyterms prompting, speaker diarization and a medical mode add-on
• Unlimited concurrent streams and Python and JavaScript SDKs

Strengths and weaknesses

• Strength: Lowest published streaming rate in this list at $0.15 per hour
• Strength: Median latency is documented, which helps voice agent planning
• Strength: No commitment on pay as you go
• Weakness: The fastest realtime model costs $0.45 per hour, three times Universal-Streaming
• Weakness: Universal-Streaming is English only; the multilingual version covers 6 languages
• Weakness: Add-ons such as keyterms and diarization are billed per hour on top

Key facts

• Batch languages: 99 on Universal-2
• Realtime latency: About 150 ms p50 (Universal-3.6 Pro Realtime)
• Concurrency: Unlimited concurrent streams

Where it ranks

AssemblyAI is No. 1 in our Best Speech-to-Text APIs verdict.