
Meta platforms are reportedly planning to launch the latest variant of its large language model (LLM) later this month, after delaying it twice.
A report from The Information suggested that Meta could delay the launch of Llama 4 once again, citing two people familiar with the matter.
Leading technology firms have been investing in the evolving AI infrastructure following the immense success of OpenAI’s ChatGPT, which revolutionised the technological realm.
Related: Meta to build data centre in US state to support AI growth
The report suggested that one of the reasons for the pushback is that the Llama 4 model is currently under development.
In addition, Meta was concerned that Llama 4 is less capable as compared to OpenAI’s models in conducting human-like voice conversations.
Llama 4 is likely to borrow particular technical aspects from DeepSeek, a Chinese AI firm, which created waves in the AI landscape.
With up to one variant scheduled to employ machine-learning techniques called a mixture of expert method, that trains separate parts of models for certain tasks, making them experts in those areas.
Meta has considered launching Llama 4 through Meta AI first and then as open-source software later, the report further added.
In 2024, Meta launched its expensive free Llama 3 AI model, which is able to converse in eight languages, write enhanced-quality computer code and solve more complex tasks than previous variants.
Related: Meta signs multi-year deal as UFC's ‘first fan technology partner’