Meta Prepares New AI Models Under Wang Leadership

Meta Platforms is preparing to release its first new artificial intelligence models developed under the leadership of Alexandr Wang, with plans to make some versions available under an open source license, according to company insiders. The initiative, part of Meta's Superintelligence project launched in June 2025, reflects Chief Executive Mark Zuckerberg's vision to advance AI for widespread personal use. Wang, former CEO of Scale AI, joined Meta earlier this year through a $15 billion acquisition of his company.

The upcoming models mark a shift in Meta's AI strategy, adopting a hybrid approach that balances openness with proprietary controls. While earlier Llama models were fully open source, the new releases will keep the largest and most advanced components closed to address safety risks and protect competitive edges. Officials indicated that the rollout will occur in phases, starting with assessments of potential hazards before broader distribution. This cautious method aims to allow developers access while safeguarding core capabilities, such as those involving cybersecurity or multimedia generation.

Meta's focus differs from rivals like OpenAI and Anthropic, which...