Muse Spark 1.3

Meta's multimodal reasoning model designed for long-running agentic, multi-agent, and coding workflows with a 1-million-token context window.

Strengths

Handles extended reasoning across long task sequences and can maintain state across multiple interactions and tool calls.

Best for

Agentic workflows, multi-step coding tasks, and applications that need to track information over extended conversations.

Limitations

A reasoning-focused model; for cost-sensitive experimentation or simpler tasks, the Contributor tier or Glimmer models offer tradeoffs in capability for lower overhead.