Models & Labs
Meta releases Muse Spark Open, a massive open-weight coding model
Hugging Face Bloghigh confidence
Why it matters
- →Sets a new benchmark for the scale of US open-weight models available to the public.
- →Provides specialized infrastructure for running massive MoE models via NVFP4 quantization on Blackwell GPUs.
- →Offers a multimodal foundation specifically tuned for complex agentic coding and debugging workflows.
Meta has released Muse Spark Open, an open-weight multimodal model with 816 billion total parameters and 42 billion active parameters per token. The model is optimized for agentic coding tasks, supporting inputs of text, code, images, audio, and video within a 1 million token context window. Two checkpoints are available: a standard BF16 version requiring distributed inference across multiple nodes, and an NVFP4 quantized version designed for single-node NVIDIA Blackwell deployment. The release includes full model weights, tokenizer, and integration recipes for vLLM and SGLang.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
More from Hugging Face Blog
More in Models & Labs
Models & Labsmodels
llama.cpp b12029 adds CUDA 13 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally supporting NVIDIA's latest CUDA 13 stack alongside AMD's ROCm 10.0. The inclusion of native builds for Linux arm64 with Snapdragon chips, covering CPU, Adreno GPU, and Hexagon NPU, signals a serious push into mobile edge deployment. While Apple Silicon KleidiAI is disabled in this specific build, the broader hardware coverage means developers no longer need to compile from source for most modern accelerators. It’s less about new features and more about removing friction for anyone running local models on bleeding-edge or diverse hardware.
llama.cpp Releases
© The Verge AIModels & Labsother
