# Meta Opens Its First Paid Model API With Muse Spark 1.1

The million-token agentic model prices at 1.25 and 4.25 dollars per million input and output tokens, putting Meta directly into the coding-agent market against Anthropic and OpenAI.

- Published: 2026-07-19T05:30:22.062Z
- Canonical: https://polylog.news/ai/2026-07-19/meta-opens-its-first-paid-model-api-with-muse-spark-1-1
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Meta AI](https://ai.meta.com/blog/introducing-muse-spark-meta-model-api/), [CNBC](https://www.cnbc.com/2026/07/09/meta-jumps-into-ai-coding-market-to-chase-anthropic-and-openai.html)

Meta on July 9 introduced [Muse Spark 1.1](https://ai.meta.com/blog/introducing-muse-spark-meta-model-api/) alongside a public preview of its new Meta Model API, the first time outside developers can pay to build on a Meta frontier model. The model, from Meta Superintelligence Labs, is a multimodal reasoning system with a one-million-token context window, tuned for tool and computer use, coding, and multimodal understanding.

  The move is a strategy shift as much as a product. Meta established its reputation on downloadable Llama weights. A metered API at 1.25 dollars per million input tokens and 4.25 per million output tokens, with 20 dollars in starter credits, puts Meta on the same commercial basis as the closed labs it once positioned itself against. [CNBC framed](https://www.cnbc.com/2026/07/09/meta-jumps-into-ai-coding-market-to-chase-anthropic-and-openai.html) the launch as Meta pursuing Anthropic and OpenAI in coding specifically.

  On Chamath Palihapitiya's price scale, Meta's list price sits well below the frontier US labs and well above the lowest Chinese prices, which positions Muse Spark as a mid-price agentic option rather than a premium reasoning model. The coding claims come from the vendor and await independent benchmark reproduction.

## What this means

The channel is distribution and revenue. Meta is following the same move away from open weights toward metered APIs that the other Western labs made, pursuing the coding-agent budget where usage and revenue concentrate. The exposed party is the open-weight developer ecosystem that relied on Meta as its main supplier and is now pushed toward Chinese open weights or paid US APIs.

## What to watch

- Independent coding benchmark results for Muse Spark 1.1 against Claude and GPT-class models, which would show whether the mid-tier price reflects mid-tier capability.
- Whether Meta keeps shipping downloadable weights in parallel, since a full shift to paid-only would confirm the industry-wide move to close the weights.