aiqbench analysis

Meta's release of Muse Spark Open is an open-weight mixture-of-experts model with 816B total parameters, 42B active per token, and a 1M-token context window, and the article claims it leads open models on the MCP-Atlas benchmark and runs on two M5 Ultra Mac Studios. For practitioners, the key points are the architecture (sparse activation keeps inference cost closer to a 42B dense model) and the deployability claim, which should be verified against independent benchmarks and hardware reports before being taken at face value.