What is Meta’s Muse Spark Open and what can it do?
An open-weight model is an AI model whose core components — the numerical parameters learned during its training, known as its weights — are released publicly. Anyone can download the model, run it on their computer, study how it works, and adapt it to their needs, subject to the model’s licence and usage terms.
Meta has unveiled Muse Spark Open, an open-weight AI model designed specifically for coding and agentic tasks, including complex software engineering and multi-step reasoning. This model is a development of Meta's Muse Spark model, enhanced with additional safety measures post-training to mitigate potential misuse.
According to Meta, Muse Spark Open can process screenshots, video clips, and documents in real time, setting it apart from traditional models that rely on fixed, pre-written steps. It's touted as the most capable open-weight model to date, capable of running on a desktop using two Apple M5 Ultra Mac Studios.
The model is available through various API platforms such as Baseten, Fireworks AI, and Groq, and can be downloaded from platforms like Amazon SageMaker AI, Google Cloud Model Garden, Hugging Face, and Microsoft Foundry Models. It is compatible with inference engines like SGLang, vLLM, Hugging Face Transformers, and Apple MLX, and is optimized for use with developer tools and hardware from Apple, Intel, and Nvidia AMD.
Developers can customize Muse Spark Open using PyTorch, Meta's own machine learning framework. The model is released under the Muse Model License, which imposes limited commercial terms to encourage widespread use and adaptation by developers.
Written by urgent.news from Economic Times Tech's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
