What happened
In early July 2026, reporting indicated Meta's Superintelligence Labs is shifting away from the openly released Llama approach toward mostly closed, proprietary frontier models. Leaders have said the most capable systems may be held back to manage safety risks and protect their designs. A next-generation model has faced delays after early versions lagged rivals on coding, reasoning, and writing.
The context
Meta's open releases seeded much of today's open-source AI ecosystem, so a pivot to closed models would reshape it.
Why this moves the Board: Release execution · Materiality 4 · T3 Reported → weighed for Meta at the next refresh.