AssemblyAI vs Deepgram

AssemblyAI is ahead of Deepgram on both datasets together, the Pipecat Dataset, time to first text and final-text delay. The Ocular Dataset is too close to call.

Last updated Methodology by Dileep Chagam

Who is better at what

Each row sets AssemblyAI's best model against Deepgram's best on that measure. For costs, see our breakdown of AssemblyAI pricing.

AssemblyAIDeepgram
  • Overall WER2.41%AssemblyAI STT 3.7 Realtime Agent3.78%Deepgram Nova-3
  • Pipecat Dataset WER1.68%AssemblyAI STT 3.7 Realtime Agent3.46%Deepgram Nova-3
  • Ocular Dataset WER3.81%AssemblyAI STT 3.7 Realtime Agent4.18%Deepgram Flux English
  • Time to first text485msAssemblyAI Universal 3.6 Pro791msDeepgram Flux English
  • Final-text delay47msAssemblyAI STT 3.7 Realtime Agent103msDeepgram Nova-3

AssemblyAI vs Deepgram FAQ

  • AssemblyAI is ahead, 1.79 points lower. The best AssemblyAI result is AssemblyAI STT 3.7 Realtime Agent at 1.68%, and the best Deepgram result is Deepgram Nova-3 at 3.46%, on the Pipecat Dataset.

  • AssemblyAI is faster, 56ms sooner. The best AssemblyAI result is AssemblyAI STT 3.7 Realtime Agent at 47ms, and the best Deepgram result is Deepgram Nova-3 at 103ms, on final-text delay.

  • AssemblyAI: AssemblyAI STT 3.7 Realtime Agent, AssemblyAI Universal 3.5 Pro, AssemblyAI Universal 3.6 Pro. Deepgram: Deepgram Flux English, Deepgram Flux Multilingual, Deepgram Nova-3. Every model ran the same tests on Converse-STT.

Test your voice agent the same way.

Cekura runs these simulated calls on your agent and scores every one.