Voice AI BenchmarksBy

PROVIDERLAST 30 DAYS

official resource

AssemblyAI voice AI models and benchmarks

AssemblyAI builds managed speech recognition APIs.

Measured models
2
STT
Best STT model#8 / 24
146ms
Universal 3.5 Pro

Overview

AssemblyAI builds and serves its own endpoints, with multilingual, diarization and keyterm capabilities.

Every model below is measured daily on the same fixed inputs and ranked against the full field, never blended into a company score.

Model lineup

Speech-to-Text
Universal Streaming, Universal 3.5 Pro

Speech-to-Text

Full STT dashboard

Ranked on Time to Final Segment against 24 measured models.

Time to Final Segmentms · lower is better · AssemblyAI models markedEvery measured STT model on Time to Final Segment, with AssemblyAI's models highlighted.
  1. #1STT RT v564 ms
  2. #2Parakeet TDT 0.6B v370 ms
  3. #3STT 183 ms
  4. #4Nova 399 ms
  5. #5Nova 2101 ms
  6. #6Ink 2108 ms
  7. #7Scribe v2 Realtime120 ms
  8. #8Universal 3.5 Pro146 ms
leaders plus AssemblyAI models · 16 other models in the full table
Word Error Rate% · lower is better · AssemblyAI models markedEvery measured STT model on Word Error Rate, with AssemblyAI's models highlighted.
  1. #1Universal 3.5 Pro3.2%
  2. #2resonant-13.4%
  3. #3Chirp 34.1%
  4. #4Enhanced4.3%
  5. #5GPT-4o Transcribe4.7%
  6. #6Grok STT4.8%
  7. #7STT 14.8%
  8. #21Universal Streaming7.0%
leaders plus AssemblyAI models · 20 other models in the full table
Time to First Tokenms · lower is better · AssemblyAI models markedEvery measured STT model on Time to First Token, with AssemblyAI's models highlighted.
  1. #1Universal 3.5 Pro1030 ms
  2. #2Flux1089 ms
  3. #3Flux Multilingual1175 ms
  4. #4Parakeet TDT 0.6B v31210 ms
  5. #5Whisper Large v31241 ms
  6. #6Nova 31434 ms
  7. #7Nova 21436 ms
  8. #10Universal Streaming1510 ms
leaders plus AssemblyAI models · 16 other models in the full table
TTFS vs WER — AssemblyAI vs every measured modelEach point is one measured STT model · AssemblyAI models highlighted · 30-day averagesTime to Final Segment against Word Error Rate for every measured STT model, with AssemblyAI's models highlighted.
STT RT v5 — 64 ms, 5.8%Parakeet TDT 0.6B v3 — 70 ms, 11.0%STT 1 — 83 ms, 4.8%Nova 3 — 99 ms, 6.3%Nova 2 — 101 ms, 7.9%Ink 2 — 108 ms, 4.9%Scribe v2 Realtime — 120 ms, 5.5%Default via Azure — 155 ms, 5.3%Whisper Large v3 — 180 ms, 8.2%Velma 2 STT Streaming — 191 ms, 6.4%Grok STT — 199 ms, 4.8%Pulse — 205 ms, 5.3%Default via Speechmatics — 223 ms, 5.5%Default via Gradium — 249 ms, 10.0%resonant-1 — 288 ms, 3.4%Enhanced — 341 ms, 4.3%Voxtral Mini Transcribe Realtime 2602 — 356 ms, 6.2%GPT Realtime Whisper — 559 ms, 5.2%GPT-4o mini Transcribe — 623 ms, 5.1%Solaria 1 — 680 ms, 7.1%GPT-4o Transcribe — 742 ms, 4.7%Chirp 2 — 811 ms, 5.0%Chirp 3 — 813 ms, 4.1%Universal 3.5 Pro — 146 ms, 3.2%
Benchmarked models with their Time to Final Segment over the last 30 days.
ModelHostTTFSRank
Universal StreamingAssemblyAI——
Universal 3.5 ProAssemblyAI146 ms8th of 24

Limits of this comparison

Coval tracks its accuracy-oriented Universal model and its Universal Streaming endpoint separately so a product mode does not become a single provider score.

  • Coval's measurements do not cover every AssemblyAI audio-intelligence feature, language or asynchronous workflow.

Official resources

Same datasets, prompts and metric definitions for every model, measured by Coval’s open-source runner. Full methodology on the overview.

Evaluate your own voice agent

Use Coval to test your production configuration, prompts and calls—not only the public benchmark endpoints.

Book a Demo