
What happened
ai& launched Meta's first open-weight Muse model, Muse Spark Open (meta/muse-spark-open), on its ai& Inference platform, callable via an OpenAI-compatible API or playground.
Why it matters
The model runs on ai&'s infrastructure inside Japan, and inputs and outputs are not used to train the model, giving domestic users a locally hosted option for this Meta release.
WHO IT HITSJapanese developers and businesses building AI applications can now access Meta's new open-weight model through a domestic inference provider, with data processed inside Japan and not used for training.
Summaries like this, every morning.
ai&, a Japanese AI company focused on inference, has moved quickly to host Meta's newly released open-weight model. The company states a policy of making new models available domestically as soon as possible, often on the same day, and Muse Spark Open is the latest example. By running the model on its own data centers in Japan and committing not to use inputs or outputs for training, ai& aims to address data residency and privacy concerns that can matter to Japanese businesses. The model itself is a native multimodal MoE with 816B total parameters and 42B active parameters, designed for coding and long-running agent tasks, and Meta positions it as the most capable open-weight model from the US. For ai& Inference users, the addition expands the set of models available through an OpenAI-compatible API, though the body does not specify pricing details beyond usage-based billing.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Bay AI Inc. launched with $50 million — a $37 million Series A led by Battery Ventures and a $13 million seed…

Kora Systems, led by Cyanogen co-founder Kirt McMaster, unveiled a Linux-based desktop OS that runs AI models…

Google unveiled TPU 8tx, built with AMD for confidential training and inference, and TPU 8i Max, which combine…

Anthropic will offer 825 million Class A shares at $70 to $78 each, raising $57.75 billion to $64.35 billion…

IPROS announced five free 20-minute sessions at AI-Industry World 2026, covering Microsoft 365 Copilot, busine…

Stratistics MRC forecasts the AI accelerator and AI semiconductor market will reach $285 billion by 2034, up f…
