Meta AI: What Is Muse Spark? And What Happened to Llama?

Why It Matters

By moving to a proprietary, multimodal model, Meta aims to stay competitive in the fast‑evolving generative‑AI market and monetize AI directly within its consumer ecosystem. The change also signals a broader industry trend away from open‑source dominance toward tightly controlled, revenue‑generating AI services.

Key Takeaways

• Muse Spark is Meta's first closed‑weight multimodal model
• Model ranks seventh on Artificial Analysis, ahead of DeepSeek V4 Pro
• Llama models sidelined after falling behind Chinese open models
• Muse Spark currently limited to Meta AI apps, no public API

Pulse Analysis

Meta’s introduction of Muse Spark represents a decisive turn from its earlier open‑source strategy. After years of championing the Llama family, the company now bundles a proprietary, multimodal model into its consumer products, positioning AI as a built‑in feature rather than a downloadable asset. This move mirrors the broader shift among tech giants toward closed‑weight models that can be tightly integrated, monetized, and protected from rapid replication by rivals.

Muse Spark distinguishes itself with three operating modes—Instant, Thinking, and Contemplating—that enable everything from quick answers to deep, parallel reasoning across sub‑agents. Trained with input from over a thousand physicians, the model excels at visual tasks such as health‑related image analysis and calorie estimation, though it still lags in niche problem‑solving like Sudoku. Its ranking at seventh place on the Artificial Analysis leaderboard places it ahead of many open models, yet it remains behind industry leaders such as GPT‑5.5 and Claude Opus, highlighting both progress and the competitive gap Meta must close.

For Meta’s business, Muse Spark could become a differentiator across its ecosystem, powering the Meta AI chat, Instagram, WhatsApp, and future hardware like Ray‑Bans. However, the lack of a public API limits enterprise adoption and may cede ground to OpenAI and Anthropic, whose agentic coding tools dominate the high‑value subscription market. Meta’s success will hinge on expanding Muse Spark’s capabilities, opening broader developer access, and translating its consumer‑centric AI into sustainable revenue streams.