Meta AI Muse Spark: A guide to Meta's AI models

Meta has repositioned its AI strategy around Muse Spark, a proprietary multimodal reasoning model from Meta Superintelligence Labs that supports visual analysis, web search, tool use, and agentic workflows, replacing Llama as the company’s flagship model. Although Muse Spark ranks competitively with leading frontier systems, Meta retains some open-weight offerings through Muse Glimmer, a 30-billion-parameter model designed for locally run agents, alongside Muse Image for image generation. The broader Muse product line includes the Muse personal-agent app, which operates through a dedicated cloud virtual machine and can complete ongoing online tasks under oversight from a safety agent; Muse Code, a cross-platform terminal coding assistant with lower-priced subscription plans; and the forthcoming Muse Charm voice-based physical AI device. Muse Spark powers Meta AI across its social platforms and is available through apps and an API, including a discounted Contributor tier that allows Meta to train on user inputs and outputs. The transition followed Llama 4’s weaker-than-expected performance against Chinese open models such as DeepSeek and Qwen, benchmark controversy, and a major 2025 restructuring that created Meta Superintelligence Labs, signaling Meta’s move from open-model leadership toward competition in proprietary AI and personal-agent services.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.