According to reports, Meta Platforms is preparing to release open-source versions of its upcoming AI models. The company is currently developing new proprietary models, internally called “Avocado”, a large language model, and “Mango”, a multimodal generator. These models are expected to launch later this year, while the open-source versions will come afterward.

The open versions won’t include all the features of the closed models and may be limited or restricted for safety and resource reasons. This move shows that Meta wants to make its AI technology more widely accessible, while still balancing openness with the company’s competitive and security concerns.