Meta Releases Muse Spark Open, an Open-Weight AI Model Built to Run on Macs
The 816-billion-parameter model is open to developers now, but running it yourself takes very high-end hardware.
Meta launched Muse Spark Open on Thursday, October 15, according to TechCrunch. Meta calls it the most capable open-weight model in the United States. Open-weight means the trained model files are shared, so others can run and build on them.
TechCrunch reports the model has 816 billion parameters and a 1 million token context window, which is how much text it can consider at once. It handles text and images, and is built for coding and agent tasks. Meta released it with Muse Code, its AI coding terminal, and says it used the pair for bug fixes, new features and a full build of a Mastodon client.
Running it on your own machine takes serious hardware. TechCrunch cites Engadget saying it needs at least 1 TB of RAM, and says it works best on a Mac Studio with the M5 Ultra chip. Meta is also offering it through several API platforms, though the article names none and gives no pricing or licence terms.
Most small businesses should not try to host this themselves. The useful part is that open models give software makers cheaper options, so watch for tools you already use to add features or lower prices. If you need private AI, ask your vendors whether they can use open-weight models.
- Meta launches an open-weight model that runs on Macs (TechCrunch)
Is AI recommending your business?
Our AI visibility audit checks what ChatGPT, Gemini and Google's AI say about you, and includes a 1:1 review of the results.
See the audits