FrontierThe story, in brief
Cohere Transcribe Arabic is an open-source model built for Arabic's toughest transcription problems
Cohere's 2B Arabic transcription model just outperformed Whisper on dialects. Here's why that matters for 400M Arabic speakers.
The KeyNews take
Why it matters
Cohere is challenging OpenAI's Whisper dominance in non-English speech recognition by releasing a specialized, open-source model optimized for Arabic's linguistic complexity (dialects, code-switching, bilingual scenarios). This signals a shift toward language-specific model development and open-source alternatives in underserved markets.
The key facts
5 to knowCohere Transcribe Arabic: 2-billion-parameter model
Outperforms Whisper and OmniASR on dialects, code-switching, bilingual Arabic-English speech
Open-source release under Apache 2.0 license
Available on Hugging Face
Published July 7, 2026
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Cohere has released Transcribe Arabic, an open-source model for Arabic speech recognition that the company says outperforms Whisper and OmniASR on dialects, code-switching, and bilingual Arabic-English speech. The 2-billion-parameter model is available on Hugging Face under the Apache 2.0 license.