Granite Speech 5.0 2B

Use it to transcribe speech or translate speech to and from English, including keyword-biased transcription and translation.

Availability

Released October 20, 2026; available on Hugging Face

Pricing model

Not established

Deployment

Hugging Face Inference Endpoint, or local use with Transformers, vLLM, llama.cpp, and mlx-audio

Privacy

Not established

Licence

Apache 2.0

• Supports multilingual ASR for English, French, German, Spanish, Portuguese, and Japanese. ^[1]

  Granite Speech 5.0 2B is a compact and efficient speech-language model, specifically designed for multilingual automatic speech recognition (ASR) and bidirectional automatic speech translation (AST) for English, French, German, Spanish, Portuguese and Japanese.

• Supports automatic speech translation to and from English for the supported languages, plus English-to-Italian and English-to-Mandarin. ^[1]

  to and from English for the same languages, plus English-to-Italian and English-to-Mandarin.

• Supports keyword-biased ASR and speech translation with keyword lists of up to 2,000. ^[1]

  Granite Speech 5.0 2B is a compact and efficient speech-language model, specifically designed for multilingual automatic speech recognition (ASR) and bidirectional automatic speech translation (AST) for English, French, German, Spanish, Portuguese and Japanese.

• The model is available on Hugging Face and has a hosted Inference Endpoint. ^[1]

  A [Hugging Face Inference Endpoint](https://huggingface.co/docs/inference-endpoints) serving this model with vLLM is available, so you can try it without any local setup.

• The model is supported in Transformers, vLLM, llama.cpp, and mlx-audio. ^[1]
• The model is intended for enterprise applications involving speech-input processing. ^[1]

  The model is intended to be used in enterprise applications that involve processing of speech inputs.

• A speech model adds multilingual transcription and translation Daily brief 2026-10-26