TITLE: Meta Develops New AI Models for Images, Videos and Text in 2026

Meta Advances AI Development with New Models

Meta Platforms Inc. is developing two new artificial intelligence models, one focused on images and videos and another on text, with plans for release in the first half of 2026. The initiatives are part of the company's superintelligence lab, led by Alexandr Wang, co-founder of Scale AI.

The image and video model, internally codenamed 'Mango,' and the text-based model, known as 'Avocado,' were detailed in an internal question-and-answer session at Meta on Thursday. Wang, along with Meta's chief product officer Chris Cox, outlined the roadmap during the meeting. The company aims to position these models as significant advancements in AI capabilities.

Wang emphasized that the text model will prioritize improvements in coding functionality. Additionally, Meta is exploring 'world models' that can process visual information, reason, plan and execute actions without requiring training on every conceivable scenario. These developments signal Meta's push toward more autonomous and versatile AI systems.

Challenges in the AI Race

Meta has struggled to keep pace with competitors such as OpenAI, Anthropic and Google in the rapidly evolving AI landscape. The company's AI division underwent substantial restructurings this year, including leadership shifts and efforts to recruit top researchers from rival firms.

The Meta Superintelligence Labs (MSL), established to drive these innovations, has seen high-profile hires but also notable exits. Several researchers who joined MSL have since departed the company. In a further blow, Meta's chief AI scientist, Yann LeCun, announced last month that he is leaving to launch his own startup focused on 'world model' technologies.

These departures come amid broader talent competition in the AI sector, where companies offer lucrative incentives to attract expertise. Meta's recruitment strategy has aimed to bolster its internal capabilities, but retention remains a challenge.

Strategic Reliance on Existing Platforms

Meta currently lacks a standalone AI product that has achieved market dominance. Its AI assistant, integrated into the company's vast social networks, benefits from billions of users across platforms like Facebook, Instagram and WhatsApp. The assistant appears prominently in app search bars, driving usage through this embedded accessibility rather than independent appeal.

This integration has sustained engagement, but analysts note that upcoming projects from MSL carry significant expectations. Success with 'Mango' and 'Avocado' could help Meta close the gap with leaders in generative AI, where tools like OpenAI's ChatGPT and Google's Gemini have set benchmarks for innovation and adoption.

The company's CEO, Mark Zuckerberg, has publicly committed substantial resources to AI, including massive investments in computing infrastructure. Meta's approach emphasizes open-source elements in some projects, contrasting with the more proprietary strategies of some rivals.

Broader Implications for AI Development

The announcement underscores the intensifying focus on multimodal AI systems that combine text, images and videos. Such models could transform applications in social media, content creation and enterprise tools, areas where Meta has deep expertise.

As AI regulations evolve globally, Meta's developments occur against a backdrop of increasing scrutiny. Recent legislative moves, such as New York's RAISE Act, aim to address AI safety, while international bodies discuss standards for emerging technologies.

Meta's 2026 timeline positions it to potentially influence the next wave of AI adoption. However, the success of these models will depend on overcoming technical hurdles and stabilizing its AI leadership team. Industry observers will watch closely as Meta navigates these challenges to deliver on its ambitious roadmap.