news 
Meta’s next Llama models may have upgraded voice features
Meta, the parent company of Facebook and Instagram, is reportedly planning to introduce improved voice features with its upcoming Llama 4 model. This development comes as a significant upgrade to the previous iterations, enabling users to interrupt the model mid-speech.
According to an article in Financial Times, Meta has been focusing on creating a system that allows users to seamlessly interact with the AI model through voice commands, similar to OpenAI’s Voice Mode for ChatGPT and Google’s Gemini Live experience. This advancement is expected to revolutionize the way we engage with AI technology, making it more intuitive and user-friendly.
During a recent Morgan Stanley conference, Meta Chief Product Officer Chris Cox emphasized that Llama 4 will be an “omni” model capable of natively interpreting and outputting speech, text, and other data formats. This indicates that the company is shifting its focus towards creating AI models that can adapt to various input methods.
The sudden rush to upgrade Llama development stems from the impressive performance of DeepSeek’s open AI models, which have outperformed Meta’s own offerings in several benchmarks. In response, Meta has allegedly set up war rooms dedicated to analyzing how DeepSeek achieved such remarkable results at a lower cost and deploying their models efficiently.
With the upgraded voice features, Meta aims to make Llama 4 an even more valuable tool for developers, content creators, and everyday users alike.
Source: https://techcrunch.com/2025/03/07/metas-next-llama-models-may-have-upgraded-voice-features/
