Moonshot has released the weights for Kimi K3, a massive MoE model with 2.8 trillion parameters, alongside a technical report. This model, designed for long-horizon coding and large context tasks, is noted for its advanced architecture but is currently impractical for home use due to its size. Concurrently, Thinking Machines has launched Inkling-Small, a smaller MoE model with 276 billion total parameters and 12 billion active parameters, offering multimodal capabilities and a 1 million token context window, though it also requires substantial hardware resources for effective operation. AI
IMPACT The release of Kimi K3, a 2.8T parameter MoE model, pushes the boundaries of open-weight models, while Inkling-Small offers multimodal capabilities in a more accessible, though still demanding, package.
RANK_REASON Cluster contains announcements of new large language models (Kimi K3, Inkling-Small) with associated technical details and weight releases from their respective labs (Moonshot, Thinking Machines). [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
