Muse Spark 1.1 is Meta's multimodal reasoning model built around agentic workflows rather than plain chat. OpenRouter describes it as a system designed to orchestrate multi-agent setups, acting as either a planning main agent or a delegated subagent, and to generalize zero-shot to new tools, MCP servers, and custom skills. That positioning is reinforced by support for structured output, parallel function calling, built-in search with citations, and configurable reasoning effort, which together make it suitable for pipelines where the model has to plan steps, call external functions, and hand off work to other agents. Meta also reports strong results on real-world coding across large codebases, computer-use workflows, and visual-to-code generation, suggesting the model is tuned more for sustained, tool-heavy work than for casual conversation.
Independent evaluation lines up with that agentic focus. NanoGPT's review of third-party testing, citing Artificial Analysis, gives Muse Spark 1.1 an Intelligence Index score of 51, up from 43 for the original Muse Spark and roughly on par with GPT-5.4 xhigh, GPT-5.6 Luna at max, and GLM-5.2 at max in the same framework, with particular strength in coding, difficult analysis, and large-document work. The same review flags weak spots in instant chat, the hardest autonomous agent tasks, and time-to-first-token latency, which matters for interactive use cases. Practically, the model fits teams building agent stacks that need long context, mixed text and media inputs, and reliable tool use, while leaner chat or latency-sensitive applications may want a lighter alternative.
