For three years, Meta was the undisputed champion of open-source AI. The Llama model family — from Llama 1 in 2023 to Llama 3.1 in 2024 and Llama 4 in 2025 — was the foundation of an entire ecosystem. Startups, researchers, and enterprises around the world built on Meta’s open models, and Mark Zuckerberg personally evangelized the open-source AI philosophy in blog posts, interviews, and shareholder letters. Then, in early 2026, reports emerged that Meta was developing its most powerful models under internal codenames — “Avocado” (a frontier LLM) and “Mango” (a multimedia generator) — and that these models might not be released as open-source.
The Shift Nobody Expected
The reports, first surfaced by industry journalists and subsequently confirmed by sources within Meta, indicate that the company is moving toward a “hybrid” approach under Chief AI Officer Alexandr Wang. Smaller, general-purpose models would continue to be released openly, maintaining Meta’s developer ecosystem and community goodwill. But the most powerful frontier models — those with capabilities that raise safety concerns or represent significant competitive advantages — would remain proprietary, available only through Meta’s own products and APIs.
This represents a fundamental strategic reversal. Zuckerberg’s original argument for open-source AI was compellingly simple: Meta makes money from ads and social media, not from selling AI models. Open-sourcing models commoditizes the technology layer, attracts developers to Meta’s ecosystem, and delivers free improvements from the global research community. The strategy worked brilliantly — Llama became the most widely used open model family in the world.
Why the Change
Several factors appear to be driving Meta’s shift:
- Safety concerns. As AI models become more capable, the risks of open release increase. A model that can generate highly persuasive misinformation, create realistic deepfakes, or assist with harmful activities poses risks that don’t exist with less capable models. The Anthropic Mythos situation — a model restricted because of its vulnerability-finding capabilities — illustrates that frontier capabilities can create genuine safety dilemmas.
- Competitive intelligence. Open-sourcing a model means your competitors can study your architecture, training methodology, and capabilities in detail. With Chinese AI labs reportedly using open models as baselines for their own development (and allegedly using distillation to harvest intelligence from API-accessible models), the competitive cost of openness has increased.
- Revenue pressure. Meta’s AI spending has been enormous — tens of billions of dollars per year on compute infrastructure alone. As investors demand returns, the case for keeping the most valuable models proprietary and monetizing them directly becomes harder to resist.
- The security breach factor. Meta reportedly paused collaboration with a data contractor following a security breach, highlighting the operational risks of broad access to frontier capabilities.
What This Means for the Open-Source AI Ecosystem
Meta’s potential retreat from fully open frontier models is significant because of the ecosystem that depends on Llama. Thousands of companies, research institutions, and individual developers have built products, papers, and careers on the assumption that Meta would continue releasing its best models openly. If the most capable models go proprietary, the open-source ecosystem doesn’t die — Google’s Gemma, Mistral, and others remain committed — but it loses its most prolific contributor at the frontier.
The philosophical implications are equally significant. Meta’s open-source releases were the strongest argument that AI development could be both open and responsible. If even Meta concludes that the most capable models need to be restricted, it strengthens the position of OpenAI and Anthropic, who have always argued that frontier models require controlled release.
The Broader Pattern
Meta’s shift reflects a broader maturation of the AI industry. In the early phase, open-source was a competitive weapon — a way for Meta to undermine OpenAI’s business model and attract developers. As the technology matures and the stakes increase, the calculus changes. The most capable models become strategic assets that are too valuable to give away and potentially too dangerous to release without safeguards.
Whether Meta’s hybrid approach is the right balance — or whether it’s simply a step toward full proprietary development under the cover of continued open-source releases of less capable models — remains to be seen. But the era of Meta giving away its absolute best is likely over. And the open-source AI community will need to recalibrate its expectations accordingly.