Holy crap, Mark Zuckerberg just posted on social platform X for the first time in three years, announcing the official launch of the latest-gen intelligent agent and coding large model, Muse Spark 1.1. This model is now fully integrated with Meta AI and the newly launched Meta Model API, signaling that Meta is aggressively reshaping the developer ecosystem for cutting-edge large language models and automated agents with a highly competitive low-pricing strategy. From the technical specs and performance metrics, Muse Spark 1.1 is designed for the automated execution of long-sequence complex tasks. The model features a 1 million token-long context window for reasoning and supports breaking down core engineering tasks and distributing them in parallel to multiple sub-agents for collaborative work. Additionally, the model has undergone targeted reinforcement training for computer use interfaces across desktop, mobile, and browser environments. According to the official benchmark data, Muse Spark 1.1 has surpassed or come close to the mainstream performance of competitors like Gemini 3.1 Pro, Opus 4.8, and GPT-5.5 in multiple agent performance and engineering coding tests. Third-party developers have revealed the pricing model for the Meta Model API, which adopts an extremely aggressive low-cost strategy: Input costs are approximately $1.25 per million tokens, cached input costs $0.15, and output costs $4.25.