Meta's Mark Zuck announces new AI model Mango to compete with Sundar Google, Sam OpenAI

Mark Zuckerberg is pushing Meta straight back into the AI fight with Mango, a new image and video model that sits beside Meta’s next text model, Avocado, all built to take on Google’s new Gemini tools and OpenAI’s image systems.

Meta wants both models in the first half of 2026, according to Alexandr Wang, the company’s chief AI officer, during an internal Q&A with Chris Cox.

Alexandr also said Meta is starting work on early world models, which means AI systems that learn by taking in visual data from their environment.

Meta changed its whole AI structure in the summer.The company hired Alexandr to run a division called Meta Superintelligence Labs. Zuck then went on a hiring spree and grabbed more than twenty researchers from OpenAI.

He built a group of over fifty people with deep AI skills. Meta is trying to load up on specialists because image generation is now one of the main battlegrounds between big model companies.

Google and OpenAI models are going to be tough to compete with

Meta released a video generator called Vibes in September. The tool was made with Midjourney. A few days later, OpenAI rolled out its own product, Sora, to keep up. The back-and-forth showed how fast each company reacts.

Google had already kicked off another wave earlier in the year with Nano Banana, which boosted monthly users of Gemini from 450 million in July to more than 650 million by late October. That jump placed even more pressure on the other players.

The tension rose again in November when Google launched the third version of Gemini.Right after that, people inside OpenAI said Sam Altman declared a “code red” to push the company to win back top scores on model tests.OpenAI then released a new version of ChatGPT Images.

Sam later met journalists and said image creation is now one of the biggest reasons users return to AI apps, calling it a “sticky” feature. The message was clear: the image fight is not slowing down.

Google announced Gemini 3 Flash on Wednesday. The model is built to run faster and at lower cost, and it is meant for wide use. It carries many of the reasoning skills found in Gemini 3 Pro but in a smaller system.

The whole plan is simple. Instead of keeping the best tools locked behind enterprise plans, Google wants these models inside mass-market apps.

Alphabet CEO Sundar Pichai said, “With this release, Gemini 3’s next-generation intelligence is now rolling out to everyone across our products including Gemini app + AI Mode in Search. Devs can build with it in the Gemini API, Google AI Studio, Gemini CLI, and Google Antigravity and enterprises can get it in Vertex AI and Gemini Enterprise.”

Get $50 free to trade crypto when you sign up to Bybit now