Meta starts charging for AI with Muse Spark 1.1 and a new developer API

For years, Meta's AI strategy was to give its models away and make money elsewhere. That era officially ended on July 9, when Meta Superintelligence Labs released Muse Spark 1.1 alongside a new Meta Model API — the first time the company has charged businesses for access to one of its models.

What Muse Spark 1.1 is

Muse Spark is the model family that replaced Meta's long-running Llama line after the company reorganized its AI efforts under Alexandr Wang's Superintelligence Labs. Version 1.1 is a multimodal reasoning model built for agentic work: it handles text and images, uses tools, writes code, and operates software.

The most distinctive feature is how it handles big jobs. Muse Spark 1.1 is trained to orchestrate multi-agent systems — it gathers context, makes a plan, and then delegates pieces of the work to parallel sub-agents to finish faster. Combined with a context window of one million tokens, it is designed for the kind of sprawling, many-file, many-step projects that trip up single-threaded assistants.

The business model shift

The new Meta Model API launched in public preview with a paid tier priced at $1.25 per million input tokens and $4.25 per million output tokens. That positions Muse Spark 1.1 as one of the cheapest capable agentic models available — below Grok 4.5 at $2/$6 and far below OpenAI's flagship pricing.

The strategic read is straightforward: Meta watched OpenAI, Anthropic, and Google build multi-billion-dollar API businesses and decided free distribution alone was leaving too much on the table. It is now competing directly for the same developer dollars, and it is doing so the way challengers usually do — on price.

A launch in heavy traffic

Muse Spark 1.1 arrived the same day OpenAI opened its GPT-5.6 family to the public, and the same week as Grok 4.5 — capping the most crowded stretch of major model releases the industry has seen. Four frontier labs shipped or opened flagship-tier models within roughly ten days of each other.

Why it matters for small business

The week's real story is that top-tier AI is becoming a buyer's market. Meta entering the paid API game adds yet another credible, aggressively priced option, which pressures every provider your software vendors build on. Practically: if you pay for any AI-powered tool, expect better pricing or higher usage limits to trickle down over the next quarter, and do not be shy about asking for them. And if you have been putting off a custom AI project because quotes came in high six months ago, get fresh quotes — the underlying model costs have dropped substantially, and a competent developer now has at least four cheap, capable models to choose from.