Meta Charges for AI for the First Time — Muse Spark 1.1 and the Meta Model API

by Geth Author

For years, Meta's strategy for AI was simple: release powerful open-weight models for free and let the community do the rest. On July 9, 2026, that changed. Meta Superintelligence Labs announced Muse Spark 1.1 — the company's first model available through a paid commercial API — placing it in direct competition with OpenAI and Anthropic in the developer market for the first time.

What Muse Spark 1.1 Is

Muse Spark 1.1 is a multimodal reasoning model built for agentic tasks. That means it goes beyond answering questions: it can use tools, control software through computer use, write and run code, and coordinate long multi-step workflows. The model supports a 1-million-token context window, large enough to ingest entire codebases or lengthy document sets in a single call.

It's the second model from Meta Superintelligence Labs — the division Meta assembled earlier this year by recruiting heavily from Google DeepMind. The original Muse Spark debuted in April; the 1.1 update delivers substantial gains in tool use, coding accuracy, and multimodal understanding across image, text, and structured data inputs.

Pricing and API Access

The Meta Model API launched in public preview for US developers on July 9, priced at $1.25 per million input tokens and $4.25 per million output tokens, with $20 in free credits at signup.

Those numbers are deliberately competitive. Grok 4.5 launched from xAI just one day earlier at comparable price points; GPT-5.6 and Sonnet 5 tiers run higher for similar capability levels. Meta is entering this market on price as much as performance — a strategy that echoes how it approached the open-source model space, but with a revenue model attached this time.

Why This Marks a Turning Point

Meta's Llama family transformed the open-source AI ecosystem. Llama 3 and its successors run on millions of devices, servers, and fine-tuned deployments worldwide. But open-weight models don't generate direct revenue, and Meta has watched OpenAI and Anthropic build dominant enterprise businesses while it gave its technology away for free.

Muse Spark 1.1's commercial API signals a deliberate strategic shift. More than that, its emphasis on agentic capabilities — computer use, tool orchestration, autonomous multi-step reasoning — puts it in the category where enterprise spending is growing fastest. Companies building internal AI agents, coding pipelines, and document automation workflows represent a market worth hundreds of millions in monthly API spend, and Meta wants a share of it.

The US-only public preview limits immediate global adoption, but the signal is unmistakable: Meta is no longer content to be the open-source alternative. Whether Muse Spark 1.1 earns its place on production infrastructure will depend on how it performs against real workloads in coming weeks — but the competitive pressure it adds to OpenAI and Anthropic's pricing is already real, and immediate.

The week of July 7–11 saw three frontier labs — xAI, Meta, and OpenAI — all making aggressive API pricing moves within roughly 48 hours. The era of premium AI API pricing may be ending faster than anyone predicted.