NEWS · MODELS · #2364
IBM updates Granite Speech suite with granite-speech-5.0-2b and two revised models
IBM released three updates to its Granite Speech collection: a new speech-language model, granite-speech-5.0-2b, and revised granite-speech-5.0-470m-turboctc and granite-speech-4.1-2b-plus. The 5.0-2b model (trained on ~390,000 hours of public audio) expands multilingual transcription and translation and improves English ASR robustness; the Turbo CTC 470M revision adds multilingual support, native punctuation, and faster runtime (~19,870 RTFx); the 4.1-2b-plus revision improves speaker attribution, timestamps, long-form WER, and keyword biasing. All models ship under Apache 2.0 with integrations for Transformers, vLLM, llama.cpp and mlx-audio.
KEY POINTS
- IBM released three updates to its Granite Speech collection: a new speech-language model, granite-speech-5.0-2b, and revised granite-speech-5.0-470m-turboctc and granite-speech-4.1-2b-plus.
- The 5.0-2b model (trained on ~390,000 hours of public audio) expands multilingual transcription and translation and improves English ASR robustness; the Turbo CTC 470M revision adds multilingual support, native punctuation, and faster runtime (~19,870 RTFx); the 4.1-2b-plus revision improves speaker attribution, timestamps, long-form WER, and keyword biasing.
- All models ship under Apache 2.0 with integrations for Transformers, vLLM, llama.cpp and mlx-audio.
WHY IT MATTERS
This matters because IBM’s updated open-weight speech models improve accuracy, speed, multilingual support and speaker attribution for enterprise transcription/translation use cases, and are broadly usable via popular ML runtimes under an Apache 2.0 license.