Meta’s 2026 story is a strategic pivot. The company that made open-source AI mainstream with Llama spent this year reorganizing around Meta Superintelligence Labs, led by chief AI officer Alexandr Wang, and shifting toward proprietary models — starting with a new “Muse” series — rather than freely released ones. It debuted those models in hardware you can actually buy, including a new wave of AI glasses starting around $299, and it is spending on the order of $100 billion-plus this year to build the data centers behind it all.
If you use Meta’s apps or are eyeing its glasses, the practical takeaway is this: Meta AI is getting more capable and more deeply embedded in Instagram, WhatsApp, Facebook, and wearables, but the era of Meta handing its best models to everyone for free may be ending. Below is what changed, why it matters, and what to watch. This is based on Meta’s announcements and public reporting.
What changed: Superintelligence Labs and a closed-source turn
The structural news is Meta Superintelligence Labs (MSL), the division Meta built to chase what Wang calls “personal superintelligence” — the idea of a digital “second brain” that thinks and acts on your behalf. Wang, who joined to lead Meta’s AI after Meta’s large investment in his previous company, has reorganized the effort around this goal and recruited aggressively.
The more consequential shift for the industry is philosophical. Meta reportedly moved away from openly releasing its most capable models, making the new Muse series proprietary, with only vague hopes of open-sourcing later versions. For years Meta’s open Llama models were the backbone of the open-source AI movement, giving developers and hobbyists a free, powerful base to build on. Pulling back from that is a notable reversal — and a loss for people who valued running capable AI on their own hardware. If local, private AI matters to you, Best Local AI Tools 2026: Run AI on Your Own PC covers the alternatives that remain.
Muse models and Meta AI in your apps
Meta’s first major model under the new regime is Muse Spark, positioned as the debut of its proprietary series and built in part for efficient on-device inference — including in wearables. The practical goal is a smarter Meta AI assistant across the products billions of people already use: Instagram, WhatsApp, Facebook, Messenger, and the glasses.
For a normal user, the assistant is the touchpoint. Meta AI can answer questions, generate images, help with messages, and increasingly see and hear through your glasses. It’s convenient precisely because it’s built into apps you already open. Whether it’s your favorite assistant is another question — on raw capability, the frontier is still set by OpenAI, Google, and Anthropic, and we map that race in The AI Model Race 2026: OpenAI vs Google vs Anthropic and rank assistants in Best AI Chatbots 2026: ChatGPT vs Claude vs Gemini & More.
The hardware bet: AI glasses go mainstream-ish
Meta’s clearest 2026 consumer story is wearables. It rebranded its companion app to the Meta AI app and launched a broader lineup of AI glasses — reported models starting around $299 — designed around hands-free, voice-first AI and better cameras. The glasses ship with Meta’s own model for fast on-device responses, letting you ask questions, get translations, and capture photos and video without pulling out a phone.
This is Meta’s real differentiator. No rival has a comparable combination of billions of app users and a maturing line of AI wearables. Glasses are where Meta thinks “personal superintelligence” becomes tangible — an assistant that sees what you see. It’s still early, and battery life, privacy, and genuine usefulness remain open questions, but the category is moving from novelty toward everyday gadget.
The money and the ambition
None of this is cheap. Meta’s AI-related capital spending in 2026 reportedly runs to roughly $115–135 billion, close to double the prior year, funding massive new data centers built at unusual speed. Wang has framed the payoff in grand terms — arguing superintelligence could unlock some of the biggest scientific discoveries in history within years.
That’s a bold claim, and worth reading with a skeptical eye. The gap between “personal superintelligence is coming” and today’s helpful-but-limited assistants is large. Meta’s spending also feeds the broader debate about whether the industry’s AI investment is rational or a bubble, which we examine in Is the AI Bubble Real? A Level-Headed Look. For Meta, the bet is existential: it missed owning a mobile platform, and it doesn’t want to miss the next one.
Where Meta sits versus rivals
In the frontier model race, Meta is a fast-moving challenger rather than the leader — OpenAI, Google, and Anthropic set the pace at the very top. Meta’s advantages are distribution (billions of users across its apps), hardware (glasses), and capital. Its weaknesses in 2026 include the reputational cost of retreating from open source and questions about whether its reorganized labs can consistently ship frontier-class models.
For choosing what to use day to day, Meta AI is convenient if you live in its apps, but the assistant to beat on quality is still ChatGPT, Claude, or Gemini. See Meta’s full profile in Meta AI Review 2026: Is It a Good Brand? and our wider The AI Directory.
What to watch next
- Open source or not. Watch whether Meta open-sources future Muse versions or completes the shift to closed models.
- Glasses adoption. The AI glasses line is Meta’s best shot at a new platform. Real-world uptake is the test.
- Model quality. Watch whether Muse models close the gap with the frontier leaders.
- Privacy. Always-available AI glasses raise real questions about consent and recording that regulators are watching.
FAQ
What is Meta Superintelligence Labs?
Meta Superintelligence Labs (MSL) is Meta’s AI division, formed in 2025 and led by chief AI officer Alexandr Wang, focused on building advanced AI toward what Meta calls “personal superintelligence.” It develops Meta’s models, including the new Muse series, and the AI features across Meta’s apps and hardware.
Is Meta still open-sourcing its AI models?
Reporting in 2026 indicated Meta shifted away from openly releasing its most capable models, making its new Muse series proprietary with only vague plans to open-source later versions. This reverses the approach that made Llama a foundation of the open-source AI community. Existing open models remain available, but the newest ones may not be.
What is Muse Spark?
Muse Spark is Meta’s first major model in its new proprietary Muse series, developed by Meta Superintelligence Labs and built partly for efficient on-device use, including in AI glasses. It powers a more capable Meta AI assistant across Meta’s apps and wearables.
How much are Meta’s AI glasses in 2026?
Meta launched a broader AI glasses lineup in 2026 with reported starting prices around $299, alongside higher-end display models. They emphasize hands-free, voice-first AI, translation, and improved cameras, running Meta’s own model for fast on-device responses. Prices and models vary by region and configuration.
Is Meta AI better than ChatGPT?
Not on raw capability — the frontier is still set by OpenAI, Google, and Anthropic. Meta AI’s advantage is convenience: it’s built into Instagram, WhatsApp, Facebook, and Meta’s glasses, so it’s already where you are. For the best standalone assistant, ChatGPT, Claude, or Gemini generally lead. See Best AI Chatbots 2026: ChatGPT vs Claude vs Gemini & More.
Who is Alexandr Wang?
Alexandr Wang is Meta’s chief AI officer, leading Meta Superintelligence Labs since 2025. He joined after Meta’s major investment tied to his previous company and now drives Meta’s push toward “personal superintelligence.” He has publicly argued that advanced AI could enable significant scientific breakthroughs within a few years — a bold claim worth reading skeptically.
Zen Tech Hub may earn a commission from links on this page, at no extra cost to you.
