Moonshot AI releases Kimi K3 open-weight model with 2.8 trillion parameters for public download

Kimi K3's full weights are here, but they're 'open' with a caveat: What enterprises should know

Chinese AI startup Moonshot AI, creators of the increasingly famous "Kimi K" family of powerful, open AI models, today released the full weights for its largest and most performant version yet -- Kimi K3. But enterprises evaluating the model should read the attached custom Kimi K3 usage license as carefully as the benchmark charts. VentureBeat previously covered Kimi K3 when it debuted through Moonshot's hosted API earlier this month, including its 2.8 trillion-parameter architecture, one million-token context window and frontier benchmark performance. Today's announcement completes that rollout with the release of the full model weights, a 47-page technical report documenting its training innovations and obstacles, and much of the infrastructure required to run the model independently. That includes the complete 2.8 trillion-parameter Mixture-of-Experts model, inference infrastructure, optimized attention kernels, MoE communication libraries and deployment components aimed at researchers and enterprise developers who want to self-host the system rather than consume it exclusively through an API. Moonshot is also releasing implementation support for ecosystems including vLLM and SGLang, alongside a technical report detailing the architectural innovations behind the model, including Kimi Delta Attention, Attention Residuals and Stable LatentMoE. Together, those techniques underpin what the company describes as the world's first open 3T-class model, activating 104 billion parameters from a pool of 896 experts while supporting native multimodal reasoning and a one million-token context window. While Kimi K3's license grants broad rights to developers and even enterprises to download, modify and deploy the model for commercial purposes -- a win for those seeking low-cost, frontier-level AI they can control and run offline -- it also imposes some added obligations on larger companies and AI service providers not found in traditional open-source licenses like Apache 2.0 or MIT. The key restrictions enterprises should understand Here's the text of the new Kimi K3 License in full: Permission is hereby granted, free of charge, to any person (the "Licensee") obtaining a copy of this software -- including the model weights, parameters, configuration files, inference and training code, and associated documentation (collectively, the "Software") -- to deal in the Software without restriction. This includes, without limitation, the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software; to run, deploy, fine-tune, or otherwise modify the Software and create derivative works from it; and to permit persons to whom the Software is furnished to do so, in each case subject to the following conditions: 1. The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. Licensee's use of the Software must comply with applicable laws and regulations. 2. "Model as a Service" means giving a third party access to language model inference or fine-tuning (e.g., via API) in a manner that allows such third party to exercise meaningful control over the inputs, parameters, or training data. This does not include (a) end-user products with model capabilities solely embedded within specific features or harnesses, or (b) mere relaying of requests to models hosted by others. If the Licensee or any of its affiliates operates a Model as a Service business, and the aggregate revenue of the Licensee and its affiliates exceeds 20 million US dollars (or the equivalent in other currencies) in total over any consecutive 12 months, the Licensee must enter into a separate agreement with Moonshot AI before using the Software or its derivative works for any commercial purpose. 3. If the Software (or any derivative works thereof) is used for any of the Licensee's commercial products or services that have more than 100 million monthly active users, or more than 20 million US dollars (or equivalent in other currencies) in monthly revenue, "Kimi K3" must be prominently displayed on the user interface of such product or service. 4. The requirements set forth in Sections 2 and 3 do not apply to: (a) internal use of the Software, defined as any use that does not make the Software, its outputs, or its underlying capabilities available to third parties; or (b) any use of the Software accessed through Moonshot AI's official products or certified inference partners. The most significant clause is 2., as it governs companies operating what Moonshot defines as a "Model as a Service." That wording is notable because it does not limit the revenue calculation to products built on Kimi K3. Instead, it references the aggregate revenue of the licensee and its affiliates, potentially bringing even smaller companies that are part of larger parents into the requirement that they must commercially license the model from Moonshot, pending the parent firm generates revenue of $20 million or more per year. Importantly, the revenue threshold alone does not appear to trigger the requirement. The clause applies specifically to organizations operating as a "Model as a Service" business, making that definition central to determining whether a commercial agreement is required. Therefore, simply using the model as a customer service chatbot or other public-facing backend shouldn't trigger a commercial license. However, in clause 3, the license separately requires attribution once a commercial deployment reaches significant scale: "If the Software... is used for any of the Licensee's commercial products or services that have more than 100 million monthly active users, or more than 20 million US dollars in monthly revenue, 'Kimi K3' must be prominently displayed on the user interface." For enterprise software vendors, AI copilots and consumer applications, that branding requirement may prove just as significant as the revenue threshold. Companies that typically abstract away the underlying model may instead need to disclose it directly within their products. But, if your enterprise is sticking to using Kimi K3 internally as an employee tool for information retrieval, creating spreadsheets and documents, or answering employee questions, Moonshot offers a significant carve out, stating that clauses 2 and 3 "do not apply to: (a) internal use of the Software, defined as any use that does not make the Software, its outputs, or its underlying capabilities available to third parties." Developers immediately zeroed in on the license The licensing provisions quickly became one of the dominant topics of discussion following the release of the model weights. AI researcher Nathan Lambert, previously the co-leader of the Olmo family of open models at AI startup Ai2, summarized the issue in a post on X: "Kimi K3 license. It's inspired by MIT but distinctly non-commercial, where any company making over $20M/yr must get a specific commercial deal (and display Kimi K3 if over 100M users or $20M/mo revenue)." Lambert's assessment reflected what many developers noticed as they dug into the newly published license: while Kimi K3 offers unrestricted access to the model weights for researchers, startups and many enterprises, commercial obligations become significantly more complex for larger organizations. The broader community reaction was largely positive toward the release itself. Developers praised Moonshot for publishing not only the weights but also supporting infrastructure, including attention kernels, MoE communication libraries and agent tooling, viewing the release as a significant contribution to the open-weight AI ecosystem. Others highlighted the rapid pace of ecosystem support, with inference projects such as vLLM and SGLang, along with cloud providers and infrastructure partners, moving quickly to support Kimi K3 deployments. At the same time, discussion centered on two practical caveats. One was licensing: many developers argued the model is more accurately described as open weight than fully open source, given the commercial conditions attached to larger deployments. The other was operational. At roughly 1.5 TB of model weights, Kimi K3 remains a system aimed primarily at well-resourced organizations capable of operating large-scale inference infrastructure, even as reports emerged of successful deployments on clusters of consumer RTX 5090 GPUs. Not the first to customize open AI licensing Moonshot is far from the first frontier AI developer to embrace an "open, but not completely open" licensing strategy. Meta's Llama family, for example, has long been distributed under its own community license requiring a commercial agreement for those building with the model and exceeding 700 million monthly users, rather than a traditional open-source software license. Other frontier model developers have likewise adopted bespoke licensing terms governing redistribution, commercial use or attribution. Kimi K3 follows that broader trend, albeit with a different mechanism. Rather than broadly restricting redistribution, Moonshot ties certain commercial rights to company scale. Organizations operating a Model-as-a-Service business above specified revenue thresholds must negotiate a separate commercial agreement, while the largest commercial deployments must visibly attribute Kimi K3 within their products. For enterprises, the practical implication is that "open weights" and "open source" are becoming increasingly distinct concepts. Downloading and modifying frontier models may be straightforward; understanding the legal conditions attached to commercial deployment increasingly is not. What enterprise leaders should do next For CIOs, chief AI officers and engineering leaders, the first step is to determine how the organization intends to use Kimi K3 before evaluating its technical performance. If the model will remain entirely inside the organization -- for example, supporting developers, researchers, legal teams or internal productivity workflows -- the published license appears substantially more permissive. Those deployments may qualify as internal use under Moonshot's terms, avoiding the commercial licensing provisions that apply to customer-facing AI services. Organizations planning to build products on top of Kimi K3 should take a different approach. Legal, engineering and product leaders should determine whether the planned deployment constitutes "Model as a Service" under the license, whether the company or its affiliates exceed the $20 million revenue threshold, and whether future growth could trigger the requirement to negotiate a commercial agreement with Moonshot. Companies expecting products to reach more than 100 million monthly active users -- or more than $20 million in monthly revenue -- should also evaluate the license's attribution requirement and how it fits with existing branding, contractual commitments and white-label offerings. More broadly, Kimi K3 illustrates a transition taking shape across frontier AI. As the industry's most capable models increasingly become available as downloadable weights instead of exclusively through hosted APIs, enterprises will need to evaluate licensing terms with the same rigor they apply to benchmarks, security reviews and infrastructure planning. The next competitive battleground may not simply be whether AI models are open or closed, but the increasingly nuanced legal frameworks that determine who can commercialize them, under what conditions and at what scale.

China's Moonshot just handed developers a near-frontier AI model to download

Kimi K3 is now available to download, modify, and host independently Moonshot AI has released the full weights for Kimi K3, giving developers the freedom to download, modify, fine-tune, and host the model themselves. The move comes shortly after K3's debut, when Moonshot's benchmarks placed it close to Claude Fable 5 and GPT-5.6 Sol, and even ahead in a handful of tests. K3 has 2.8 trillion parameters and a 1-million-token context window. Downloading K3 is easy, but running it is not Open weights make the trained parameters that shape a model's behavior publicly available. Companies can adapt K3 using private data, build specialized tools around it, or offer hosted versions under Moonshot's license without relying entirely on its API. K3 is still not fully open source, however, since Moonshot has not released its complete training data or every part of the training process. The model is available through Hugging Face, but it is far too large for an ordinary PC. Its mixture-of-experts design activates 104 billion parameters at a time, while the MXFP4 weights alone are expected to occupy roughly 1.4TB before accounting for runtime overhead. Moonshot says K3's new architecture delivers about 2.5 times more intelligence per unit of compute than Kimi K2. Even so, self-hosting the model will require a serious multi-GPU server setup. K3 is now caught in a US policy fight The release arrives in the middle of a growing US debate over Chinese AI and open-weight models. White House officials have accused Moonshot of distilling Anthropic's Fable model and training K3 using restricted Nvidia hardware. Moonshot has not publicly responded to those claims. Meanwhile, Nvidia CEO Jensen Huang is backing a growing push in the US to protect open-weight AI. The number of companies supporting a letter published on July 24 reportedly doubled from 25 to 50 within a day, adding names such as OpenAI, Google, AMD, Cisco, Cloudflare, GitHub, Block, and Ollama. The group is urging Washington not to place sweeping restrictions on downloadable models, arguing that open weights support innovation, cybersecurity research, and the ability to run AI on private infrastructure. Recommended Videos K3's release gives developers a capable new model to work with, but its size will limit who can actually run it. Cloud providers and larger companies are likely to be the first to offer it at scale. Amazon and Anthropic have not signed the letter, showing that major technology companies still disagree over how freely advanced AI models should be released.

Moonshot AI releases Kimi K3 open-weight model for download

The Beijing-based startup is making its 2.8-trillion-parameter model available to developers to download, modify, and host Moonshot AI will make the weights of its Kimi K3 model available for unrestricted public download on Monday, giving developers the freedom to deploy, adapt, and self-host it. Founder Yang Zhilin has said the company wants to grow its user base through openness and broader availability than competing proprietary systems in the U.S., according to Bloomberg. With 2.8 trillion parameters, Kimi K3 holds the distinction of being the largest open-weight model publicly available, according to Bloomberg. Its context window stretches to one million tokens, meaning users can feed entire large codebases or lengthy documents into a single query. A technical report covering the model's design, how it was trained, and how it performs on standard benchmarks is expected to follow the weight release. The weight release follows Kimi K3's debut last week, which sent Chinese AI competitor stocks sharply lower. Z.ai stock lost as much as 30% of its value in Hong Kong trading, MiniMax Group stock fell as much as 16%, and Alibaba stock dropped 4%. Arena blind evaluations showed developers selecting Kimi K3 ahead of leading U.S. models on front-end coding tasks. The model placed below Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol in overall rankings but outperformed both companies' prior-generation models across coding and general agent evaluations. Since K3 launched, Moonshot's daily revenue has grown by a factor of at least six. The company reached $300 million in annual recurring revenue in June, up from $200 million in April, and is seeking a new funding round at a $50 billion valuation ahead of a potential Hong Kong initial public offering as soon as this year. The release arrives amid scrutiny from Washington. White House Office of Science and Technology Policy Director Michael Kratsios accused Moonshot last week of training K3 using banned Nvidia $NVDA chips and conducting large-scale distillation against U.S. models, including Anthropic's Fable model. Moonshot has not responded to requests for comment on the allegations. Monday's weight release is expected to benefit cloud computing providers that specialize in hosting third-party open models. None of the three major hyperscale cloud platforms -- Amazon $AMZN AWS Bedrock, Microsoft $MSFT Azure Foundry, or Google $GOOGL Vertex AI -- have integrated Kimi K3 or comparable Chinese open-weight models into their offerings, Bloomberg Intelligence analysts Mandeep Singh and William Tong noted.

What to know about Moonshot AI and its new open weight model Kimi K3

The release of the Chinese AI model Kimi K3 was a flashpoint in the AI world, sharpening the debate over whether the most capable models should be closely held by individual companies, usually U.S. tech firms, or freely and transparently distributed in the tradition of open-source software. Not only did the high-performing Kimi K3 challenge the cherished Silicon Valley notion that Western AI labs still lead their Chinese counterparts in large language models, though likely by only a hair, but it also raised the possibility that more transparent open-weight models may ultimately be easier to control, and therefore safer, than closed-weight models whose makers must guard against misuse. What is Kimi K3? Beijing-based Moonshot AI released Kimi 3 in mid-July, followed by the model's full weights on July 27. With roughly 2.8 trillion parameters, it is likely the largest open-weight model ever released. The model is natively multimodal, meaning it can process images and audio alongside text. It can also reason across a 1-million-token context window. Why did Kimi K3 take off so quickly? The Kimi K3 weights are available on Hugging Face, a widely used clearinghouse for open-weight models. Hugging Face CEO Clem Delangue said the model reached No. 1 on the platform's trending chart with more than 4,000 likes within 30 minutes of its release, the fastest growth the site had ever recorded. By the next day, the Kimi K3 repository had about 7,700 likes and tens of thousands of downloads.

Moonshot AI readies to release Kimi K3 for public download

Kimi K3 rivals OpenAI and Anthropic's latest models across various benchmarks. Moonshot is gearing up to make the world's largest open-weight AI model available for public download, after initial interest overwhelmed the company's systems. The Chinese AI start-up is due to release the Kimi K3's weights on Monday (27 June), enabling users to fine tune and run the model locally for free. Moonshot unveiled K3 earlier this month, boasting 2.8trn parameters and performances closely rivalling that of OpenAI and Anthropic's newest models. The multimodal model is designed for frontier intelligence across long-horizon coding, knowledge work and reasoning, according to its maker. At the time of its launch, Kimi said that K3 performed "competitively" with Fable 5 and "substantially outperformed" Opus-4.8, GPT-5.6 Sol and GPT-5.5. Bloomberg reported that initial interest drove daily sales at Moonshot up sixfold since K3's debut. The company temporarily paused taking on new subscribers just days after K3's launch, after an unexpected number of users swarmed to try out the new model. New subscriptions are still paused, however, interested buyers can join the waitlist. Chinese AI companies are attracting swarms of new users with models that offer similar capabilities at significantly cheaper prices, despite the many restrictions Washington placed on semiconductor exports to China over recent years. Reports show that such tactics led by the US pushed China to focus on its own chip capabilities instead. Kimi K3's launch garnered a similar reaction to DeepSeek's release early last year, which authorities in the West viewed with scepticism, floating security concerns. Despite this, today's release is expected to be profitable for cloud businesses that specialise in hosting third-party open models. Moonshot hit a $20bn valuation in May after announcing a $2bn raise led by Chinese food delivery company Meituan's VC arm Long-Z Investments, alongside Shuimu Capital, China Mobile and CPE Yuanfeng. Now, it is reportedly working to close a funding round that could value the three-year-old start-up at more than $30bn. Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news.

2.88 Trillion Parameter Kimi K3 Largest Open Model is Now Available

Kimi K3, the latest release from Moonshot AI, has officially launched as the largest open source AI model to date, boasting an impressive 2.88 trillion parameters. Built on a mixture of experts (MoE) architecture, Kimi K3 is designed to dynamically allocate computational resources, achieving 2.5 times the intelligence per compute unit compared to earlier systems. Universe of AI explores how this model's standout features, including a 1-million-token context window and native multimodal capabilities, position it as a versatile option for tasks ranging from natural language processing to advanced visual analysis. The release also emphasizes accessibility, with licensing terms carefully crafted to balance inclusivity and responsible use. Dive into this breakdown to uncover how Kimi K3's architecture enables it to excel in processing large-scale, intricate inputs and multimodal reasoning. You'll also gain insight into the open source resources provided by Moonshot AI, such as high-performance attention kernels and sandbox environments for safe experimentation. Additionally, the guide highlights Kimi K3's performance benchmarks, which rival proprietary models like GPT 5.6 Soulmax and discusses its implications for the future of open AI innovation. What Makes Kimi K3 Stand Out? Kimi K3 is built on a mixture of experts (MoE) architecture, boasting an impressive 2.88 trillion parameters. This design enables the model to allocate computational resources dynamically, achieving 2.5 times the intelligence per compute unit compared to earlier systems. Its standout features include: * A 1-million-token context window, allowing the model to process vast and intricate inputs with ease. * Native multimodal capabilities, seamlessly integrating text and visual data for enhanced versatility. * Optimized efficiency, making it suitable for tasks ranging from natural language processing to advanced visual analysis. These innovations position Kimi K3 as a highly adaptable and powerful tool, capable of addressing complex challenges across diverse fields. Open source Tools for Developers Moonshot AI has complemented the release of Kimi K3 with a robust suite of open source tools designed to empower developers and researchers. These resources include: * High-performance attention kernels and communication libraries to streamline training and inference workflows. * Sandbox environments for safe experimentation and fine-tuning, allowing users to customize the model for specific applications. By providing these tools, Moonshot AI underscores its commitment to fostering a collaborative ecosystem, encouraging innovation and allowing the global AI community to push boundaries. Advance your skills in Kimi K3 by reading more of our detailed content. Licensing: Balancing Accessibility and Responsibility The licensing framework for Kimi K3 is carefully crafted to balance accessibility with responsible use. Inspired by the MIT license, it incorporates specific conditions to ensure ethical deployment. Key aspects include: * Free access for individuals and small organizations, promoting inclusivity and widespread access of AI technology. * Commercial entities with over $20 million in annual revenue or more than 100 million users must obtain a commercial license. * Attribution to Moonshot AI is mandatory for commercial use, making sure proper recognition of the model's origins. This approach ensures that innovative AI remains accessible while encouraging large-scale commercial users to contribute to the model's ongoing development and sustainability. Performance Benchmarks: A New Standard Kimi K3 has undergone rigorous testing against industry benchmarks, delivering results that place it among the most advanced AI models available. Highlights of its performance include: * Results that rival top proprietary models such as Fable 5 Max and GPT 5.6 Soulmax, demonstrating its competitive edge. * Exceptional performance in tasks requiring large context windows and multimodal reasoning, showcasing its versatility and advanced capabilities. These benchmarks highlight Kimi K3's potential to drive advancements in both research and practical applications, setting a new standard for open source AI. Open source Innovation: A Growing Trend The release of Kimi K3 reflects a broader shift in the AI industry toward open source collaboration. This trend is particularly prominent in regions like China, where open access to innovative models and tools is accelerating innovation. By making advanced AI systems available to the global community, initiatives like Kimi K3 are fostering a more competitive and collaborative environment. This approach contrasts with the more closed strategies often adopted by Western labs, emphasizing the importance of transparency and shared progress in driving technological advancements. Looking Ahead: The Future of AI The AI landscape is evolving rapidly, with new models on the horizon promising even greater capabilities. Upcoming releases such as GPT-6, Deepseek 4G, and GLM 5.3/5.5 are expected to push the boundaries of AI performance further. Additionally, specialist models like Microsoft's MAI Cyber 1 Flash are gaining traction, offering tailored solutions for niche applications. These developments signal a future where AI systems are not only more powerful but also increasingly specialized and adaptable, catering to a wide range of industries and use cases. The Competitive AI Landscape The global race for AI leadership is intensifying, driven by both established players and emerging innovators. Companies like SpaceX and Meta are introducing novel architectures and applications, raising the bar for performance and utility. This competitive environment offers unprecedented opportunities for developers and researchers to use innovative technologies, contribute to new advancements and shape the next wave of AI breakthroughs. A Milestone in Open AI The release of Kimi K3 represents a significant milestone in the evolution of open source AI. With its new architecture, extensive capabilities and accessible tools, it sets a new benchmark for what open AI systems can achieve. As the global AI landscape continues to evolve, models like Kimi K3 will play a crucial role in shaping the future of technology and innovation. Whether you're a developer, researcher, or industry leader, this is an opportunity to engage with and contribute to the fantastic potential of AI. Media Credit: Universe of AI Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.