While much of the AI industry has focused on building bigger data centers and chasing top talent, Mark Zuckerberg is taking a different approach: undercutting everyone on price.
A New Paid Tier for Developers
Meta has introduced Muse Spark 1.1, a new AI model that comes with a paid developer tier — the company’s latest move in its ongoing competition with OpenAI, Anthropic, and Google. According to reports, the new Meta Model API will be priced at roughly a quarter of what competing AI labs currently charge. Zuckerberg reportedly told Bloomberg that pricing among some rival labs has become excessive, positioning Meta’s lower rates as a direct challenge to the industry’s cost structure.
The move marks a notable shift for Meta, which has spent years distributing its AI models for free rather than monetizing them directly. That strategy is now evolving as the company looks to generate returns on its massive AI infrastructure spending, having reportedly made cuts elsewhere to fund the buildout.
Betting Big on In-House Models
Zuckerberg has framed control over Meta’s own AI models as essential to his broader ambitions. He has argued that building proprietary models — rather than depending on outside AI providers — is necessary if Meta wants to embed AI products across its platforms and reach billions of users worldwide. The push is also tied to his stated goal of developing superintelligence, a long-term bet that has driven much of Meta’s recent AI spending.
By pairing aggressive infrastructure investment with a low-cost pricing strategy, Meta appears to be aiming to make its AI tools more accessible to developers while still competing at the frontier of AI capability — a bid to combine scale with affordability in a market where rivals have leaned toward premium pricing.