META CRASHES THE AI CODING PARTY WITH ITS FIRST EVER PAID MODEL — AND ZUCKERBERG BREAKS A THREE-YEAR SILENCE ON X TO ANNOUNCE IT Meta Superintelligence Labs launched Muse Spark 1.1 on July 9, and the move marks a historic first: it is the first AI model Meta has ever charged money for. Priced at $1.25 per million input tokens and $4.25 per million output tokens, the multimodal reasoning model targets the increasingly crowded agentic coding market dominated by OpenAI and Anthropic. It can handle images, video, and PDFs, runs a one-million-token context window, and ships with built-in computer use, parallel tool calling, and structured output. On benchmarks, Meta says Spark 1.1 edges out models in its price class on agentic tasks, tool use, and frontend design. The launch was significant enough to pull Mark Zuckerberg back to X, a platform he had not posted on in three years. In a brief post, Zuckerberg called Spark “a strong agentic and coding model at a very low price,” positioning it squarely against Claude and GPT. What makes this notable is not just the model specs but the business decision behind it: Meta, which has distributed AI freely for years as a way to build developer goodwill and erode the moats of rivals, is now asking the market to pay. That is a fundamental shift in strategy, and it signals that Meta’s Superintelligence Labs is no longer content to win on volume alone. It wants to compete on quality at a price that undercuts the incumbents. Keywords: Meta Muse Spark 1.1, AI coding model, Meta AI paid model, agentic coding AI