Meta's cutting-edge AI team has reportedly unveiled its initial key models internally this month, according to the company's CTO. This development comes as Meta's new artificial intelligence lab, the Meta Superintelligence Labs, formed last year, has been under scrutiny for its performance. The team's AI models, which are still in their early stages, have shown significant promise, according to the CTO. The models include a text AI model codenamed Avocado and an image and video-focused model codenamed Mango, both expected to launch in the first quarter. The company's efforts are being closely watched as CEO Mark Zuckerberg aims to strengthen its AI leadership and compete in the highly competitive technology sector. Meta's Llama 4 model, which faced criticism for its performance, is part of this strategic push. The CTO acknowledges that the technology is not yet fully developed, emphasizing the extensive work required post-training to ensure the models are usable internally and by consumers. Despite the challenges, Meta is optimistic about its future, with the CTO predicting that 2026 and 2027 will see consumer AI trends solidify, leading to more advanced and useful models. This optimism is further supported by the company's recent marketing efforts, such as the AI-equipped Ray-Ban Display glasses, which have paused international expansion to focus on U.S. orders.
