Avencera Get notified

Choose a model

Pick a model for the audio you have.

Different recordings need different models. Find yours below, with its price an hour. With or without speaker labels, every model costs less an hour than Microsoft Azure, ElevenLabs, AssemblyAI, OpenAI, Gladia and Speechmatics.

Kinds of audio

On every chart, top left is best: fewer errors and cheaper. The dashed line marks the fewest errors each price buys.

  • General audio

    A mix of recordings, or audio that fits none of the kinds below

    Qwen3-ASR $0.09 / hour$0.10 / hour with speaker labels

    More accurate than OpenAI and Gladia

    1. Microsoft Azure 5.02% errors errors,$0.18 an hour
    2. ElevenLabs 5.24% errors errors,$0.22 an hour
    3. AssemblyAI 5.39% errors errors,$0.21 an hour
    4. AvenceraQwen3-ASR, recommended 5.40% errors errors,$0.09 an hour
    5. AvenceraGranite Speech 5.68% errors errors,$0.08 an hour
    6. OpenAI 5.96% errors errors,$0.27 an hour
    7. Gladia 5.99% errors errors,$0.168 an hour
    8. AvenceraParakeet v2 6.02% errors errors,$0.06 an hour
    9. AvenceraParakeet v3 6.18% errors errors,$0.07 an hour

  • Meetings

    Team calls, board meetings and stand-ups recorded in a room

    Granite Speech $0.08 / hour$0.09 / hour with speaker labels

    More accurate than ElevenLabs, OpenAI, Gladia and AssemblyAI

    1. AvenceraGranite Speech, recommended 6.96% errors errors,$0.08 an hour
    2. Microsoft Azure 7.05% errors errors,$0.18 an hour
    3. ElevenLabs 8.19% errors errors,$0.22 an hour
    4. AvenceraQwen3-ASR 8.31% errors errors,$0.09 an hour
    5. OpenAI 8.95% errors errors,$0.27 an hour
    6. AvenceraParakeet v2 9.09% errors errors,$0.06 an hour
    7. AvenceraParakeet v3 9.42% errors errors,$0.07 an hour
    8. Gladia 9.89% errors errors,$0.168 an hour
    9. AssemblyAI 10.63% errors errors,$0.21 an hour

  • Calls and interviews

    Sales and support calls, interviews and user research

    Qwen3-ASR $0.09 / hour$0.10 / hour with speaker labels

    More accurate than Gladia, Microsoft Azure, AssemblyAI, OpenAI and ElevenLabs

    1. AvenceraQwen3-ASR, recommended 12.14% errors errors,$0.09 an hour
    2. Gladia 12.44% errors errors,$0.168 an hour
    3. Microsoft Azure 12.71% errors errors,$0.18 an hour
    4. AssemblyAI 12.80% errors errors,$0.21 an hour
    5. OpenAI 12.90% errors errors,$0.27 an hour
    6. AvenceraParakeet v3 13.44% errors errors,$0.07 an hour
    7. AvenceraParakeet v2 13.58% errors errors,$0.06 an hour
    8. ElevenLabs 13.66% errors errors,$0.22 an hour
    9. AvenceraGranite Speech 13.79% errors errors,$0.08 an hour

  • Podcasts and video

    Podcasts, YouTube videos and webinars

    Qwen3-ASR $0.09 / hour$0.10 / hour with speaker labels

    More accurate than Microsoft Azure, AssemblyAI, ElevenLabs and OpenAI

    1. AvenceraQwen3-ASR, recommended 7.22% errors errors,$0.09 an hour
    2. Gladia 7.32% errors errors,$0.168 an hour
    3. Microsoft Azure 7.48% errors errors,$0.18 an hour
    4. AssemblyAI 7.62% errors errors,$0.21 an hour
    5. ElevenLabs 7.63% errors errors,$0.22 an hour
    6. OpenAI 7.76% errors errors,$0.27 an hour
    7. AvenceraParakeet v3 7.99% errors errors,$0.07 an hour
    8. AvenceraGranite Speech 8.14% errors errors,$0.08 an hour
    9. AvenceraParakeet v2 8.15% errors errors,$0.06 an hour

  • Earnings calls

    Investor calls and financial briefings

    Parakeet v3 $0.07 / hour$0.08 / hour with speaker labels

    More accurate than AssemblyAI and OpenAI

    1. ElevenLabs 4.80% errors errors,$0.22 an hour
    2. Microsoft Azure 5.20% errors errors,$0.18 an hour
    3. AvenceraQwen3-ASR 5.84% errors errors,$0.09 an hour
    4. AvenceraParakeet v3, recommended 5.85% errors errors,$0.07 an hour
    5. Gladia 5.94% errors errors,$0.168 an hour
    6. AssemblyAI 6.05% errors errors,$0.21 an hour
    7. AvenceraGranite Speech 6.22% errors errors,$0.08 an hour
    8. OpenAI 6.61% errors errors,$0.27 an hour
    9. AvenceraParakeet v2 6.76% errors errors,$0.06 an hour

  • Lectures and audiobooks

    Recorded courses, narration and other read-aloud audio

    Granite Speech $0.08 / hour$0.09 / hour with speaker labels

    More accurate than ElevenLabs, Microsoft Azure, Gladia and OpenAI

    1. AssemblyAI 2.11% errors errors,$0.21 an hour
    2. AvenceraGranite Speech, recommended 2.21% errors errors,$0.08 an hour
    3. ElevenLabs 2.33% errors errors,$0.22 an hour
    4. Microsoft Azure 2.71% errors errors,$0.18 an hour
    5. AvenceraParakeet v2 2.73% errors errors,$0.06 an hour
    6. AvenceraQwen3-ASR 2.94% errors errors,$0.09 an hour
    7. Gladia 2.99% errors errors,$0.168 an hour
    8. AvenceraParakeet v3 3.13% errors errors,$0.07 an hour
    9. OpenAI 3.39% errors errors,$0.27 an hour

  • Speeches

    Talks, keynotes and parliamentary sessions

    Qwen3-ASR $0.09 / hour$0.10 / hour with speaker labels

    1. ElevenLabs 1.41% errors errors,$0.22 an hour
    2. Microsoft Azure 1.65% errors errors,$0.18 an hour
    3. AssemblyAI 1.92% errors errors,$0.21 an hour
    4. OpenAI 2.65% errors errors,$0.27 an hour
    5. Gladia 2.76% errors errors,$0.168 an hour
    6. AvenceraQwen3-ASR, recommended 2.85% errors errors,$0.09 an hour
    7. AvenceraParakeet v3 3.05% errors errors,$0.07 an hour
    8. AvenceraGranite Speech 3.52% errors errors,$0.08 an hour
    9. AvenceraParakeet v2 3.78% errors errors,$0.06 an hour

  • Noisy recordings

    Field recordings, phone microphones and background noise

    Qwen3-ASR $0.09 / hour$0.10 / hour with speaker labels

    More accurate than OpenAI and Gladia

    1. AssemblyAI 9.08% errors errors,$0.21 an hour
    2. ElevenLabs 9.20% errors errors,$0.22 an hour
    3. Microsoft Azure 9.90% errors errors,$0.18 an hour
    4. AvenceraQwen3-ASR, recommended 9.93% errors errors,$0.09 an hour
    5. AvenceraGranite Speech 10.22% errors errors,$0.08 an hour
    6. AvenceraParakeet v2 11.42% errors errors,$0.06 an hour
    7. AvenceraParakeet v3 11.52% errors errors,$0.07 an hour
    8. OpenAI 12.86% errors errors,$0.27 an hour
    9. Gladia 13.73% errors errors,$0.168 an hour

  • Indian English

    Calls and recordings from speakers in India

    Qwen3-ASR $0.09 / hour$0.10 / hour with speaker labels

    More accurate than OpenAI and Gladia

    1. Microsoft Azure 3.00% errors errors,$0.18 an hour
    2. ElevenLabs 3.32% errors errors,$0.22 an hour
    3. AssemblyAI 3.49% errors errors,$0.21 an hour
    4. AvenceraQwen3-ASR, recommended 3.50% errors errors,$0.09 an hour
    5. OpenAI 3.63% errors errors,$0.27 an hour
    6. Gladia 3.70% errors errors,$0.168 an hour
    7. AvenceraParakeet v2 3.89% errors errors,$0.06 an hour
    8. AvenceraGranite Speech 3.94% errors errors,$0.08 an hour
    9. AvenceraParakeet v3 4.14% errors errors,$0.07 an hour

  • European languages

    German, French, Spanish, Italian, Portuguese, Dutch and 19 more

    Parakeet v3 $0.07 / hour$0.08 / hour with speaker labels

    1. Microsoft Azure 2.61% errors errors,$0.18 an hour
    2. ElevenLabs 2.75% errors errors,$0.22 an hour
    3. AssemblyAI 3.15% errors errors,$0.21 an hour
    4. OpenAI 3.66% errors errors,$0.27 an hour
    5. Speechmatics 4.35% errors errors,$0.75 an hour
    6. Soniox 4.44% errors errors,$0.10 an hour
    7. AvenceraParakeet v3, recommended 5.25% errors errors,$0.07 an hour
    8. AvenceraQwen3-ASR 5.52% errors errors,$0.09 an hour

  • Speaker labels only

    You already have a transcript and need to know who spoke when

    Speaker labels from community-1 $0.02 / hour

    More accurate speaker labels than Speechmatics, AssemblyAI, Soniox, ElevenLabs and Deepgram

    1. pyannoteAI 16.9% errors errors,$0.125 an hour
    2. Avenceracommunity-1, recommended 23.5% errors errors,$0.02 an hour
    3. Speechmatics 32.1% errors errors,$0.75 an hour
    4. AssemblyAI 33.2% errors errors,$0.23 an hour
    5. Soniox 36.6% errors errors,$0.10 an hour
    6. ElevenLabs 39.8% errors errors,$0.22 an hour
    7. Deepgram 46.1% errors errors,$0.258 an hour

Switch models with one line

Every model works with the OpenAI SDK. Change the model name and keep the rest of your code. For speaker labels, send speaker_labels as true, a cent more an hour, and every segment comes back with its speaker.

from openai import OpenAI

client = OpenAI(base_url="https://api.avencera.com/v1", api_key="YOUR_AVENCERA_KEY")

with open("audio.mp3", "rb") as audio:
    transcript = client.audio.transcriptions.create(
        model="avencera/parakeet-tdt-0.6b-v3",
        file=audio,
        # speaker labels, a cent more an hour; leave this out for a plain transcript
        extra_body={"speaker_labels": True},
    )

for segment in transcript.segments:
    print(segment["speaker"], segment["text"])

See every price or compare providers one at a time.