Meta has announced the intention to deploy their next artificial intelligence models in early 2026. This latest move is unsurprisingly done to solidify its position in the quickly morphing AI environment. For all of this ambitious roadmap, the technology behemoth is still without a killer AI product, creating a sense of urgency in a race for development.
At an internal Q&A session held on Thursday, Meta’s Chief Product Officer Chris Cox and Chief AI Scientist Yann Lecun unveiled plans for several innovative projects. A big highlight is a major new image and video AI model, codenamed “Mango.” Its purpose is to facilitate innovative research that advances the state of visual processing. Meanwhile, Meta has been secretly working on a new text-based model with the codename “Avocado.” The company intends to improve it further with a focus on coding tasks.
Meta’s approach ultimately aims to develop more sophisticated world models that learn to interpret visual data in conjunction with reasoning, planning, and acting on their own. This method represents a paradigm change from previous generative AI models which needed to highly train on pre-set conditions. So we want to build easier to change systems. Further, these systems will be addressing multiple challenges at the same time and should not require deep training beforehand.
Yann Lecun, who has been instrumental in Meta’s AI developments, announced during the session that he will soon depart the company to start his own venture. His departure comes with a cloud of uncertainty as to where Meta’s AI efforts might be headed without his pioneering and guiding hand.
“As many of you have heard through rumors,” – Yann Lecun
Meta’s internal roadmap for AI models is a much-needed step in this current moment. These next projects are going to be ground-breaking for our small firm. How successful they proving will go a long way in determining our competitive advantage in the AI space. Meta has great hopes for “Mango” and “Avocado.” After the long AI winter before a breakthrough flagship product, the company is anxious to put its best foot forward.
Today’s announcement marks Meta’s willingness to double down on developing deeper technologies that would not only differentiate its products but potentially lead the industry in redefining them. Only time will tell how all of these initial signs will shake out — particularly with the impending change of leadership at the top.