{"slug": "moonshot-completes-kimi-k3-rollout-with-full-model-weight-release", "title": "Moonshot completes Kimi K3 rollout with full model weight release", "summary": "Moonshot AI released the full Kimi K3 model weights and technical report on July 27, 2026, making its 2.8 trillion parameter mixture-of-experts model available to developers and researchers. The model, which activates 104 billion parameters per token and supports a 1 million-token context window, achieves 2.5 times greater scaling efficiency than its predecessor Kimi K2 through architectural innovations including Kimi Delta Attention and native quantization. The weights are hosted on Hugging Face and GitHub under the Kimi K3 License.", "body_md": "# Moonshot completes Kimi K3 rollout with full model weight release\n\nMoonshot AI released the full Kimi K3 model weights and technical report, opening its 2.8 trillion parameter model to developers and researchers.\n\nMoonshot AI has released the full model weights and technical report for Kimi K3, opening its most capable artificial intelligence model to researchers and developers.\n\nReleasing the model weights and technical report of Kimi K3.\n\nKimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.\n\nNew model architecture: 2.5x the intelligence per unit of compute, not just more params.\n\nAlongside…\n\n[pic.twitter.com/Yz5uWeMbIm]— Kimi.ai (@Kimi_Moonshot)\n\n[July 27, 2026]\n\nThe company published the weights on Hugging Face and released the technical report through its official GitHub repository. Both the code and model weights are available under the Kimi K3 License.\n\nMoonshot initially introduced Kimi K3 earlier this month but said the complete weights and additional technical details would arrive by July 27. The latest release completes that rollout.\n\nKimi K3 has 2.8 trillion total parameters and activates 104 billion parameters during inference. It uses a mixture of experts architecture with 896 experts, selecting 16 for each token, and supports a context window of more than 1 million tokens.\n\nThe report also provides additional information on Kimi Delta Attention, Attention Residuals and the model’s native quantization system. Moonshot said the architectural changes deliver approximately 2.5 times greater scaling efficiency than Kimi K2.\n\nThe release allows developers to download and deploy the model through frameworks including Transformers, vLLM and SGLang. Moonshot recommends large supernode configurations due to the model’s size and infrastructure requirements.\n\n**Disclosure:** This article was edited by Estefano Gomez. For more information on how we create and review content, see our\n\n[Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/moonshot-completes-kimi-k3-rollout-with-full-model-weight-release", "canonical_source": "https://cryptobriefing.com/moonshot-completes-kimi-k3-rollout-with-full-model-weight-release/", "published_at": "2026-07-27 17:23:55+00:00", "updated_at": "2026-07-27 17:31:55.699478+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research", "ai-products", "ai-infrastructure"], "entities": ["Moonshot AI", "Kimi K3", "Hugging Face", "GitHub", "Kimi K2", "Transformers", "vLLM", "SGLang"], "alternates": {"html": "https://wpnews.pro/news/moonshot-completes-kimi-k3-rollout-with-full-model-weight-release", "markdown": "https://wpnews.pro/news/moonshot-completes-kimi-k3-rollout-with-full-model-weight-release.md", "text": "https://wpnews.pro/news/moonshot-completes-kimi-k3-rollout-with-full-model-weight-release.txt", "jsonld": "https://wpnews.pro/news/moonshot-completes-kimi-k3-rollout-with-full-model-weight-release.jsonld"}}