Meta is cosidering using OpenAI or Gogle models across its social media apps

Meta is reportedly in talks with industry competitors, including Google and OpenAI, to enhance the artificial intelligence capabilities of its social media platforms, as outlined in a report by The Information (via Reuters). The leadership team of Meta's newly established Superintelligence Labs is considering the integration of Google's Gemini to provide users with advanced conversational and text-based responses through Meta AI. Discussions are also underway regarding the potential use of OpenAI's models to bolster Meta AI and various essential AI features across the company’s suite of social media applications. Currently, Meta AI is embedded in all its platforms, including WhatsApp, Facebook, and Instagram, along with a standalone app sharing the same name. Reports indicate that Meta has already begun incorporating external AI models into some internal tools available to its employees, allowing them to utilize Anthropic's model for coding through an internal assistant. A spokesperson for Meta emphasized their comprehensive strategy to develop top-tier AI products, which encompasses creating their models, collaborating with other companies, and open-sourcing technology. This initiative is supported by significant investment, as Meta has allocated billions to build its Superintelligence Labs, co-led by former Scale AI CEO Alexandr Wang and ex-GitHub CEO Nat Friedman. CEO Mark Zuckerberg is said to be directly involved in recruiting top talent, having successfully attracted experts from OpenAI, Google, and Apple. Following some setbacks with their Llama 4 models, which struggled to compete against rival offerings from OpenAI, Google, and Anthropic, Meta executed a major internal restructuring. The company has also delayed the launch of its Behemoth model and is now reportedly focused on a new AI model expected to debut by year-end. According to insights from Business Insider, the Superintelligence Labs is in the process of developing the Llama 4.x model, internally referred to as Llama 4.5, with plans for a late-year release. During Meta's second-quarter earnings call in June, Zuckerberg mentioned the company's advancements towards the Llama 4.1 and 4.2 models, while also highlighting ongoing efforts for next-generation models aimed at pushing technological boundaries in the coming year.