Meta is developing a new generation of image, video, and text-based AI models slated for release in the first half of 2026, according to a report from The Wall Street Journal.

Both projects are part of Meta’s broader push to strengthen its AI capabilities and are being developed under Meta Superintelligence Labs (MSL), a new internal group overseen by Alexandr Wang, co-founder of Scale AI.

According to the report, Meta is working on:

The roadmap was presented during an internal Q&A session at Meta, where Alexandr Wang and Chris Cox, Meta’s Chief Product Officer, outlined the company’s plans.

Wang reportedly said Meta aims to strengthen its text model’s coding performance while also exploring world models — systems designed to understand visual information and reason, plan, and act without requiring exhaustive training across every possible scenario.

The image and video model would represent a step toward more capable multimodal AI systems that can interpret and interact with the world visually.

Meta has faced growing pressure in the competitive AI landscape, trailing rivals such as OpenAI, Anthropic, and Google.

This year, Meta’s AI division underwent significant restructuring, including leadership changes and aggressive recruitment from competing AI firms. However, the report notes that several researchers who joined Meta Superintelligence Labs have already departed.

Compounding the challenge, Meta’s longtime chief AI scientist Yann LeCun announced last month that he is leaving the company to launch his own startup.

While Meta AI, the company’s AI assistant, reaches billions of users through integration across Meta’s social platforms, the assistant’s scale is largely driven by distribution rather than a standalone breakthrough product.

As a result, the first major AI models to emerge from Meta Superintelligence Labs are expected to play a critical role in determining whether Meta can regain momentum in the rapidly evolving AI race.

Q: What models is Meta developing?
A: Meta is working on two primary systems: “Mango,” an image and video generation model, and “Avocado,” a text-based model designed to deliver stronger coding capabilities.

Q: When are the models expected to launch?
A: Meta plans to release the new models in the first half of 2026, according to an internal roadmap shared with employees.

Q: Who is leading this effort?
A: The work is being led by Alexandr Wang, head of Meta Superintelligence Labs, alongside Meta Chief Product Officer Chris Cox.

Q: What are “world models,” and why do they matter?
A: World models aim to help AI systems understand visual information and make decisions — allowing them to reason, plan, and act without needing to be trained on every possible scenario.

Q: Why is Meta under pressure in AI right now?
A: Meta has faced internal restructuring, leadership changes, and talent turnover, while competitors continue to ship high-profile AI models and products at a faster pace.

Meta’s next generation of AI models is about more than catching up in benchmarks — it’s about whether the company can translate massive distribution into meaningful AI innovation.

Unlike competitors that have built momentum around flagship AI products, Meta’s AI reach has largely been driven by embedding its AI assistant across existing social platforms. The success of models coming out of Meta Superintelligence Labs will signal whether Meta can move from scale alone to technical leadership in multimodal and reasoning-focused AI.

For developers, businesses, and creators, Meta’s progress will help shape how quickly advanced image, video, and reasoning systems become accessible at global scale — and whether Meta remains a major player in defining how AI interacts with the visual world.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.