Meta Llama 4 will let you talk to AI like never before

Meta Platforms Inc is set to enhance its voice capabilities with the upcoming version of its artificial intelligence, Llama 4, as reported by the Financial Times. This development aligns with Meta’s strategy to focus on conversational AI agents instead of traditional text-based interactions.

The introduction of improved voice features in Llama 4 is expected in the coming weeks, aiming to facilitate a two-way natural dialogue where users can interrupt the AI rather than strictly adhering to a question-and-answer format. This shift demonstrates Meta’s commitment to transforming user interaction with AI technology.

CEO Mark Zuckerberg has outlined ambitious goals for the company, designating 2025 as a pivotal year for its range of AI products, which are under development amid increased competition from companies such as OpenAI, Microsoft, and Google. To undergird these initiatives, Meta plans to invest up to $65 billion in 2025, primarily focusing this expenditure on data center infrastructure.

Meta AI approaches 600 million monthly users with Llama 3.3

Alongside advancements in Llama 4, Meta is exploring the possibility of premium subscriptions for its AI assistant, Meta AI, which would enable it to perform tasks such as booking reservations and video creation. The company is also considering implementing paid advertising or sponsored posts in the search results of Meta AI.

The chief product officer Chris Cox indicated that Llama 4 would operate as an “omni model,” allowing speech to function natively within the system, thereby eliminating the need to translate voice into text and back. He emphasized the significance of this capability at the Morgan Stanley technology, media & telecom conference, stating, “I think we are still wrapping our heads around how powerful that is.”

Meta has been actively discussing the necessary guardrails for the latest Llama model’s output while evaluating whether to lessen these restrictions. These discussions come amidst competitive pressures as rivals like OpenAI and Elon Musk’s xAI also make strides in deploying voice features, including OpenAI’s distinct personality offerings and Grok 3, which features an “unhinged mode” that responds in deliberately objectionable ways.

Last year, Meta addressed criticism regarding its Llama 2 model, which had been reluctant to answer benign queries by unveiling a more flexible version in Llama 3. The integration of voice commands is also a key feature of Meta’s Ray-Ban smart glasses, which have gained popularity among consumers, further indicating the company’s strategy to position lightweight headsets as the primary computing devices of the future.

Featured image credit: Dima Solomin/Unsplash