ModelsAgree
← All leaderboards
🎙️

Best speech-to-text API

3 models · updated 2026-06-29

The verdict

Deepgram leads — 2 of 3 models rank Deepgram the top startup.

Not unanimous: ChatGPT picks Deepgram Nova-3.

Combined ranking

  1. 1
    Deepgram—10 pts
    GPT —Claude #1Gemini #1· Fast, accurate streaming ASR with strong real-time latency and competitive pricing.
  2. 2
    AssemblyAI—7 pts
    GPT —Claude #2Gemini #3· High-accuracy models plus built-in summarization, diarization, and audio intelligence add-ons.
  3. 3
    Deepgram Nova-3—5 pts
    GPT #1Claude —Gemini —· Best real-time accuracy, latency, and developer controls.
  4. 4
    AssemblyAI Universal-2—3 pts
    GPT #3Claude —Gemini —· Excellent async transcription with summaries and diarization.
  5. 5
    ElevenLabs Scribe—2 pts
    GPT #4Claude —Gemini —· High-quality multilingual transcription with polished speech tooling.
  6. 6
    Gladia—2 pts
    GPT —Claude —Gemini #4· High-quality transcription for noisy business calls and complex code-switching.

Not ranked (incumbents): OpenAI Whisper, Google Cloud Speech-to-Text, OpenAI gpt-4o-transcribe, Microsoft Azure AI Speech

By model

ChatGPT

  1. 1.Deepgram Nova-3
  2. 2.OpenAI gpt-4o-transcribe
  3. 3.AssemblyAI Universal-2
  4. 4.ElevenLabs Scribe
  5. 5.Google Cloud Speech-to-Text

Claude

  1. 1.Deepgram
  2. 2.AssemblyAI
  3. 3.OpenAI Whisper
  4. 4.Google Cloud Speech-to-Text
  5. 5.Microsoft Azure AI Speech

Gemini

  1. 1.Deepgram
  2. 2.OpenAI Whisper
  3. 3.AssemblyAI
  4. 4.Gladia
  5. 5.Google Cloud Speech-to-Text

Tracked by ModelsAgree · rank 1 = 5 pts … rank 5 = 1 pt · re-polled continuously