{
  "id": 20606006,
  "title": "Reson8 releases Rumble-2-Flash speech model as a free open-weights download",
  "url": "https://urgent.news/2026/11/18/reson8-releases-rumble-2-flash-speech-model-as-a-free-open-weights",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-18T14:00:57.000Z",
  "source": {
    "name": "SiliconANGLE",
    "slug": "siliconangle",
    "url": "https://siliconangle.com/2026/11/18/reson8-releases-rumble-2-flash-speech-model-as-a-free-open-weights-download/"
  },
  "original_language": "en",
  "account": "Amsterdam-based speech recognition startup Resonate Labs B.V. has made its Rumble-2-Flash transcription model available for free download with open weights. Targeted at developers and researchers needing speech recognition for languages other than English, the 500 million-parameter model supports over 40 languages. Reson8 claims their models are trained on languages other than English, where accuracy drops for specialist terms like medication names. Reson8’s models are currently in production use for meeting notetakers, AI receptionists, customer support voice agents, medical transcription, and captioning. Co-founder and CEO Raoul Ritter stated that generic speech models are designed for English, resulting in more mistakes in other languages. Reson8’s Rumble-2-Flash, one of the smallest models on Hugging Face Inc.’s Open ASR Leaderboard, averages a 7.09% word error rate across seven languages, including German, French, and Dutch. English has a 5% average error rate. For lower-resource languages such as Maltese and Basque, Rumble-2-Flash has a 17.6% mean word error rate. Reson8’s models are trained to handle noisy settings, including phone-line audio. Unlike competitors, Reson8’s service does not retain any audio or transcripts, and maintains ISO 27001 certification along with compliance with the EU’s General Data Protection Regulation and AI Act. Available under an Apache 2.0 license, Rumble-2-Flash can be accessed through Reson8’s API for a price of €0.01 ($0.001) per hour of audio.",
  "summary": "Amsterdam-based speech recognition startup Resonate Labs B.V. today released its Rumble-2-Flash transcription model as a free download with open weights. Designed for fast transcription of prerecorded audio, the 500 million-parameter model supports more than 40 languages. Reson8 is pitching the release at developers and researchers who need speech recognition built for languages other than…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}