Kimi K3：开源权重模型的升级

2024年7月16日星期四，Moonshot AI 发布了他们最新的旗舰模型 Kimi K3： https://www.kimi.com/blog/kimi-k3。K3 是一个具有 2.8 万亿参数的 MoE 模型，其权重将于 7 月 27 日发布。本文大部分内容是对生态系统现状的反思，假设 Moonshot 会遵守权重发布日期的承诺。这是一种对均衡状态的更极端的看法，如果情况是中国有类似强大的但封闭的模型（即 K3 永远不会发布），许多结果最终会处于一个中间状态。

关键事实是，无论是从开放到封闭还是从美国到中国，模型性能差距已经从争论中的 6-9 个月缩短到了更短的时间，比如 3-5 个月。

从发布材料来看，K3 是一个真正的前沿模型。自 DeepSeek R1 以来，它将是开放模型最接近前沿的模型：https://www.interconnects.ai/p/deepseek-r1-recipe-for-o1。DeepSeek R1 是另一种情况。那是一个中国实验室非常快速地转向推理模型，并比许多美国公司更快地发布的例子。Kimi K3 则是一个中国实验室在已知领域进行规模化执行的例子：数据、算法、架构、工具、环境等。

Kimi K3 在 Vals AI 指数中总体排名第 2：https://x.com/ValsAI/status/2077834614668988586，在 Artificial Analysis 的 Intelligence Index 总体排名第 3：https://x.com/ArtificialAnlys/status/2077832874183860404 （仅次于 Claude Fable 和 GPT-5.6 Sol Max，但价格更低：https://x.com/ArtificialAnlys/status/2077832885021835289?s=20），在 Frontend Code Arena 中总体排名第 1：https://x.com/arena/status/2077824029126504525，还有更多令人印象深刻的结果：https://x.com/cramforce/status/2078574147333152957?s=46。Moonshot AI 正在与 Anthropic 和 OpenAI 正面竞争，而资源少得多。

这显然是迄今为止发布过的最强大的开源模型。从这个模型可以清楚看到，如果来自美国封闭前沿模型的对抗蒸馏有所贡献，也充其量只是相对较小的程度。关注过蒸馏恐慌的 AI 观察者：https://www.interconnects.ai/p/the-distillation-panic，并错误地得出结论认为中国的 AI 实验室只是通过知识产权盗窃才能生产出优秀模型的人们，将会有所醒悟——中国公司在构建模型方面与领先的美国公司一样非常出色。Moonshot AI 正在解决许多 OpenAI 或 Anthropic 人员正在解决的同样问题。我相信将会有更多关于蒸馏的讨论和压力：https://www.interconnects.ai/p/the-distillation-panic，但现在的证据已经显示，中国公司可以做的不仅仅是快速跟随：https://www.interconnects.ai/p/how-much-does-distillation-really。

在我中国之行中与 Kimi 核心团队会面：https://www.interconnects.ai/p/notes-from-inside-chinas-ai-labs，我很清楚地感受到他们拥有令人难以置信的文化，有人甚至称之为气场，并且有表达这种文化的自由——尽管是在 GPU 限制的环境下。在构建模型高度依赖规模的情况下，构建好模型的能力很大程度上仍取决于个人执行力、动机和表达能力。访问过他们之后，这个结果就不那么令人惊讶了。在访问过众多 AI 公司后，很少有公司拥有你可以立即感受到的文化。

与此同时，中国的 AI 应用趋势起步比美国要晚。因此，尽管所有中国实验室的算力远低于美国同行，但更多的算力肯定可以用于模型训练。当我开玩笑说一个 OpenAI 平均研究员可能拥有多少算力——比如几千台 H100 等效机器——Kimi 的研究员们都感到震惊。Kimi 模型的组织结构和构建方法无疑反映了这一点，但若没有大量的专有信息，很难理清这一点的具体情况。

峰值模型性能的现状大致如下：https://x.com/finbarrtimbers/status/2077816750574539161：

Moonshot AI – Kimi K3（开源权重*）

智谱（Z.ai）– GLM 5.2（开源权重）

阿里巴巴 – Qwen 3.7 Max（已宣布 3.8，也将会是开权重版本，撰写时）

看到 DeepMind 和一些其他美国巨头表现如此低，令人惊讶。在很多方面，X AI 团队值得更多的认可。以下是来自 Artificial Analysis 的视觉总结：

此次发布以及其他近期事件导致了对于开放模型与封闭模型之间平衡的最可能结果方向的重大变化。我将逐一解析。

在许多方面，这感觉像是新纪元的开始。一个竞争更多、但同时协调需求更高的时代，因为我们正向全球推广极其强大的技术。

分享：https://www.interconnects.ai/p/kimi-k3-the-open-weights-escalation?utm_source=substack&utm_medium=email&utm_content=share&action=share

许多人最近才开始关注中国的 AI 场景，所以他们可能得出结论认为公开发布模型是他们的核心策略。实际上，我认为大多数实验室的核心策略更接近 Anthropic 或 OpenAI——构建尽可能最佳的智能。多年来，我一直关注并与中国实验室互动，对于他们最初选择公开发布模型，最好的解释是出于实用性。他们需要公开发布模型以获得采用、关注和反馈（尤其是在高价值的湾区市场）。

长期以来，中国关于开放源代码 AI 的政策非常有限：https://www.chinatalk.media/p/chinas-new-ai-plan，说明了开放源 AI 的角色，以及可能的“国家层面战略”。据我所知，没有高级领导公开评论过开放源 AI。这一情况本周发生了变化，习近平在世界人工智能大会（WAIC）上发表了主题演讲，并非常直接地承诺中国 AI 生态系统的未来将面向开源和全球扩散：https://mattsheehan.substack.com/p/xi-jinpings-big-ai-speech-annotated。这一对现状的承诺，与迄今最强开权重模型的发布在同一周，是现代 AI 早期历史中的一个明显标志。

这发生在中国人工智能产业讨论了许多潜在发展路径的时期——它们会保持开放吗？它们能跟上美国实验室在规模上的步伐吗？中国有不断增长的收入市场吗？围绕这些问题，焦点一直在于中国的风险承受能力、公司变现的能力，以及任何可能让公司停止公开发布其最佳模型的密切相关原因。

通过将习近平在时间上的承诺与一个非常强大的模型联系起来，中国隐含地评论了其在发布开放权重模型方面的风险承受能力。目前来看，这是对中国体系内诸如强大网络安全能力（或生物危害）等话题感知风险的一种解读。

最简单的解释是，中国政府确实在密切关注模型可能带来的风险——可能比美国政府的直觉监管更具技术范围——如果评估出风险将会采取行动。简单的解释是，他们认为当前的前沿模型并没有产生显著风险。

与此同时，中国的经济决策者认为推广人工智能是有益的，因此他们可以在产业成熟后获利——在扩大分销之后（正如中国在汽车、太阳能、先进制造业及近期许多领域所做的那样）。

在经历了数月关于Claude Mythos模型的炒作和恐慌的美国人工智能媒体环境中，这些可能显得有些令人震惊。这种惊讶是一个极好的基础——世界对于我们在美国听到的大多数关于人工智能的叙述，并没有达成一致。

许多引导人工智能讨论的叙述者显然有压低最佳开放人工智能模型感知能力的动机。Dean Ball——他个人支持开放模型，但现在在OpenAI工作——发表了一篇广泛评论的帖子：https://x.com/deanwball/status/2078133895766114412，其中对Kimi有一些反思，他在其中谈到了开放模型。理解这一声明很重要，因为它强调了开放模型在人工智能建设的经济层面上的作用。Dean说：1：#footnote-1

开放权重模型本质上是减速的，我持续感到惊讶的是，所谓的“加速主义者”对开放权重模型如此兴奋。

解释为什么开源权重模型是一种减速主义形式对于理解即将到来的世界秩序很重要。他说得对。

从经济角度来看，开源模型对前沿实验室是减速主义的，这将减缓 AI 的净投资和资本支出部署。这是因为强大的开源权重 AI 模型大大降低了封闭实验室的利润潜力。这有两个影响。首先，AI 实验室用于未来模型再投资的利润减少。其次，市场认为这些公司的终端价值较低，因此会削弱未来的融资轮次。这两者结合在一起，会延缓最具变革性 AI 模型的时间表，但我认为这些影响不足以阻止 OpenAI 和 Anthropic 成为世界上估值最高的几家公司之一。

对我来说，这对社会而言是总体利好。由于开源权重模型是加速主义的，通过使具有一定性能水平的智能进入门槛降低，从而促进 AI 在整个经济中的扩散：https://x.com/natolambert/status/2078164535262040448。开源模型还鼓励定制化。问题是，这种类型的扩散其本质上要比前沿 AI 实验室产品慢得多，后者直接向开发者出售工具。开源模型的潜力是几乎每个企业都能用它们来构建特定领域的代理。这种经济扩散需要极长的时间！我曾将其描述为开源权重模型起步较慢，但潜在指数增长更大：https://www.interconnects.ai/p/open-and-closed-models-are-on-different。问题在于，如果封闭模型在原始能力上领先太多，这种定制能力可能会变得无效。

我认为，增加扩散和降低 AI 实验室权力集中度的结合，对 AI 转型非常积极。它给我们更多时间去解决新能力的难题，并让更多利益相关者影响发展方向——任何一家公司都很可能在安全控制世界上最重要技术方面遇到问题。

在这个世界上，对我来说，当然重要的是最好的模型仍然由美国公司制造，这将使美国能够控制技术及其价值观的发展轨迹。我也预计情况会是如此，因为美国拥有更大的资本市场，愿意投资于人工智能（以及日益增加的利润份额），但这并非理所当然。

Kimi 的发布博客中有一些技术细节，证实了支撑我们所感受到的持续模型改进的改进类型。选取其中一个例子：

Kimi K3 基于 Kimi Delta Attention (KDA) 和 Attention Residuals (AttnRes) 架构更新，这两项架构更新旨在改善信息在序列长度和模型深度间的流动。我们还扩大了 Mixture of Experts (MoE) 稀疏性，在与稳定的 LatentMoE 框架配对时，有效激活了 896 个专家中的 16 个。结合精细化的训练和数据方案，这些结构性变化相对于 Kimi K2 在整体扩展效率上提高了约 2.5 倍，使模型能够更有效地将计算力转化为智能。

训练效率的提升确实积少成多。这将为模型带来持续的、令人难以置信的飞跃。

顺便提一下，追踪这一特定创新在生态系统中的路径是一个有趣的例子。Kimi Delta Attention (KDA) 在 Kimi Linear 论文中被引入：https://arxiv.org/abs/2510.26692，它类似于用于 Olmo Hybrid 的 Gated DeltaNet：https://arxiv.org/abs/2604.03444（我在 Ai2 时的最后一个 Olmo 模型）。Qwen 的最新模型切换到了相关架构，最近的 Nemotron 模型也采用了混合架构（但仍然更接近 Mamba 而非 Gated DeltaNet）。看到像这样的新架构理念——这些理念在学术界得到了大量推进——能够如此迅速地转化为前沿规模模型，真是太棒了。Gated Delta Networks：https://arxiv.org/abs/2412.06464 于 2024 年末提出，基于 Mamba 的理念。到 2026 年中，它们已经应用于前沿模型。

我选择关注这个例子，部分原因是 Kimi 团队给模型间的创新赋予了一个很酷的数字，但主要是为了给关于中国资源效率的更广泛讨论留出空间。

越来越明显的是，中国的实验室在资本效率方面远远领先。在一个扩展规律表明智能与有效资本成正比的世界里——而有效资本可以购买计算资源、数据和人才——这可能是你们的AI行业能够拥有的最大优势。有许多可能的原因可以解释为什么会出现这种情况，比如中国研究人员薪酬较低，但在大型语言模型研究难题上更高效，但我们可能永远无法得到如此具体的原因。

自从我撰写关于中国的笔记：https://www.interconnects.ai/p/notes-from-inside-chinas-ai-labs 之后，我听到越来越多关于中国正在兴起的数据产业的信息（远远落后于Anthropic用于数据的数十亿美元预算），以及中国实验室能够获得可观的训练计算资源（通过规避出口管制）。中国公司没有同样的推理需求（直到最近，Moonshot AI不得不暂停新订阅：https://x.com/Kimi_Moonshot/status/2078855608565207130 来访问他们的K3模型——虽然API仍然可用），因此他们更多的计算资源可以用于训练。这些领域在很大程度上影响了实验室能力的真实性，但我们对它们的测量非常有限。

现实情况是，这些中国实验室筹集的资本远远低于美国AI生态系统的任何一部分。最直接的对比是OpenAI和Anthropic，它们的公开模型略好一些。其他公司，如Google和Meta，拥有商业史上最大的现金流，但在构建模型方面落后。至于美国的新兴实验室，竞争情况甚至更加激烈——Thinking Machines最近发布了他们的第一个模型Inkling：https://thinkingmachines.ai/news/introducing-inkling/，虽然很强，但还不在Kimi K3的同一等级。

这些拥有更多资源的美国公司仍然可能迎头赶上，但你需要认真权衡我们掌握的关于模型质量的公开测量数据，而不要寄希望于希望——希望往往反映的是偏见。如果中国实验室确实有潜在优势，他们可以继续利用这一优势，打造比所有竞争对手更强大的模型！许多结果都是可能的，而 K3 应该会增加大多数人认为中国可以凭借更高效的训练努力在不久的将来在 AI 能力上完全领先的概率——即使这不是你预测的最可能结果。

资本效率的一个重要贡献可能在于方法，美国实验室正在投入大量精力以戏剧性、大步伐推动前沿，而中国实验室更专注于赶超——这种赶超的成本更低。正如学生模型通常可以在蒸馏中超过教师模型（不限于中国实验室的对抗蒸馏），采用“尝试赶超”而不是“发明下一个范式”的方法同样可能带来更强大的模型。

在 Kimi K3 发布后的那个周末，在撰写本文并广泛讨论事件时，阿里巴巴宣布：https://x.com/Alibaba_Qwen/status/2078759124914098291 将很快推出一个 2.4 万亿参数的 Qwen 3.8 模型，并开放权重。历史上，阿里巴巴一直将其最大模型通过云业务仅提供 API，因此这是另一种重大氛围转变，为开放模型经济的下一篇章打开了大门。即使该模型在基准测试上落后于 Kimi K3，它也表明中国公司可能不仅在维持其开放模型策略的现状，还在进一步倾注其中。

如果 Qwen 3.8 模型很快发布，即在下一个 Gemini 模型之前，它可能会将谷歌推到拥有最智能模型实验室排行榜的第八位——这一排行榜中国一直在攀升。同时还有更多强大的中国模型即将出现的传言，DeepSeek V4 预计将从其“预览”版本升级发布。

发表评论：https://www.interconnects.ai/p/kimi-k3-the-open-weights-escalation/comments

我认为，如果今天发布Claude Mythos作为一个开放权重模型，负面结果将相对较小。这个观点有点难以持有，因为我们对公共网络安全评估了解非常有限，而且这是一个复杂的生态系统（同时因为我信任Anthropic的许多人）。我仍然坚持这个观点。风险被夸大了。

问题是，对于最强大的AI模型，这种情况并不总是存在。更强大的模型即将到来——伴随而来的是更高的风险——所以对于最好的模型来说，在几个月前以受控、封闭的方式访问，是一个非常安全的平衡。

用户可以高度控制的开放权重模型将永远存在——你无法有效禁止数字产品，特别是来自不良行为者的产品——因为AI训练证明已经比许多分析师预期的更早实现了全球访问。

尽管如此，在我写这篇文章时，美国政府仍在尝试更多措施以限制开放权重模型：https://www.interconnects.ai/p/6-months-to-live-for-open-models。最新的是来自Axios的报道：https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi：

幕后消息：一位接近政府的消息人士告诉Axios，商务部去年曾考虑将多个中国AI实验室加入其“实体清单”，这将有效切断未经许可的美国访问。

消息人士表示，国家安全局和白宫国家网络总监办公室去年还考虑发布关于中国AI实验室威胁的建议，实际上是在阻止美国公司使用其技术。

该消息人士补充，白宫曾考虑实施一项行政命令，表示美国公司只有在能够保证安全并在被破坏时承担责任的情况下，才能托管中国模型。

另一位接近政府的消息人士称，商务部去年夏天还在政府内部传阅了草案规则，利用其权力保障国内供应链，以针对中国的开源模型。

这将使美国处于一个非常不对称的状态，美国最好的模型在网络安全任务上有防护措施，但全球的行为者可以使用优秀的中国开放权重模型来测试我们的防御能力。这只是众多例子之一，表明禁止开放权重模型不仅会损害AI的自由市场，还会在短期内使整个生态系统变得不那么安全。还有其他非常糟糕的结果，比如减缓AI应用和AI研究的传播，正如我上面讨论的那样。

这些平衡状态很难维持，特别是随着AI工具加速模型的进展，但在开放与封闭之间维持这种现状非常重要。拥有一个在能力上真正处于前沿的模型——比如当Mythos宣布时——同时又是开放权重的，在我们进入未知能力领域时会带来严重风险。模型的发展速度将非常快，而且越来越难以衡量它们的整体能力。

所以我们陷入了一个必须在开放模型上小心调节的世界。不希望如此强大的模型瞬间全球传播是合理的，但与此同时，模型的制造者被激励去夸大其能力，而其竞争者则被激励去夸大风险。归根结底，这一切都依赖于仔细的测量和主动强化社会以应对风险向量。

这种政策辩论、模型发布及喧嚣反应的急转列车，从今天开始只会继续。自去年12月Claude Opus 4.5发布以来，我们一直处在快速发展的列车上，那次发布把我们带入了代理路径。做出良好决策的关键是评估能力，而不依赖于拥有最大财务利益的公司。进入AGI时代的AI治理时，其中一项必要行动（参见：https://www.interconnects.ai/p/welcome-to-the-agi-era-of-ai-governance）是采用类似“极速行动”的方式启动国家能力（以及其他独立主体），以便能够准确评估模型，并研究新出现的风险。

开放权重模型通过加速能力的扩散，在人工智能的潜在利与弊方面都是一次巨大的升级。目前，前沿语言模型的负面影响大多仍处于假设阶段，但这种情况不会永远持续。

让开放权重模型略微落后于封闭前沿模型，是我们缓解风险的自然缓冲。关键点在于，我们必须共同采取行动，在潜在伤害出现时加以减轻，无论开放权重模型落后3个月、6个月还是9个月，时间依然非常短。如果我们过于严格地监管开放权重模型，我怀疑世界上大部分人会被误导，认为我们不再需要采取行动。我们所做的只会是稍微延迟不可避免的结果——开放模型最终仍会跨越所有关键能力门槛，无论其合法性如何。

理解并从这种开放-封闭的互动中获益，必须是未来几年AI社区在所有权力和影响力部门共同采取的集体行动。

Kimi K3 是一个分水岭时刻，因为前沿开放权重模型已经变为现实。许多假设将会被验证，以了解开放权重模型的风险真正落在哪里——我怀疑其范围会比许多人预期的要窄，而许多AI的风险仍会通过封闭和“更安全”的API扩散。这些风险的评估将随着AI在我们经济中的整合加速而发展。二者不可分割，我们将继续同时获得二者。

因此，2025年是开放模型开始被更认真看待的一年——尤其当中国以明显领先的姿态跃进时：https://atomproject.ai/——人们意识到这不再是一个单极世界，仅由美国封闭的AI实验室决定发展轨迹。2026年，之前讨论过的潜在风险以及真正前沿的开放权重模型所带来的加速效应开始显现。

大量回应集中在第四点，该点将开放模型的不可避免结果与AI共产主义进行比较，而我认为这一点并不准确。具体而言，未加解释使用共产主义一词，引发了大部分的反弹。

On Thursday July 16th, Moonshot AI released their latest flagship model Kimi K3 ：https://www.kimi.com/blog/kimi-k3 . K3 is a 2.8T parameter MoE model which will have its weights released on July 27th. Much of this article follows as a reflection on the state of the ecosystem, under the assumption that Moonshot keeps their promise of the weights release date. This is a more extreme view of the equilibrium, and many of the results end up in a middle ground if the state of affairs is that China has similarly powerful, but closed models (i.e. K3 is never released).

The key fact is that either the open-to-closed or American-to-Chinese model performance gap has been reduced from the debated 6-9 months to something shorter, say 3-5 months.

From the release materials, it is clear that K3 is a true frontier model. It will be the closest open models have been to the frontier since DeepSeek R1 ：https://www.interconnects.ai/p/deepseek-r1-recipe-for-o1 . DeepSeek R1 was a different story. This was a Chinese lab being extremely quick to pivot to reasoning models and release one faster than many American companies. Kimi K3 an example of a Chinese lab executing on scaling the known areas: data, algorithms, architecture, tools, environments, etc.

Kimi K3 comes in at #2 overall on the Vals AI index ：https://x.com/ValsAI/status/2077834614668988586 , #3 overall on Artificial Analysis’s Intelligence Index ：https://x.com/ArtificialAnlys/status/2077832874183860404 (only beaten by Claude Fable and GPT-5.6 Sol Max while being cheaper ：https://x.com/ArtificialAnlys/status/2077832885021835289?s=20 ), #1 overall in Frontend Code Arena ：https://x.com/arena/status/2077824029126504525 , and more impressive results ：https://x.com/cramforce/status/2078574147333152957?s=46 . Moonshot AI is going toe to toe with Anthropic and OpenAI with far, far fewer resources.

It is clearly the strongest open model ever released. It should be clear looking at this model that if adversarial distillation from the closed frontier models in the U.S. contributed, it is at most to a relatively small degree. AI observers who followed the distillation panic ：https://www.interconnects.ai/p/the-distillation-panic and came away with the wrong conclusion that Chinese AI labs are only producing good models due to IP theft are in for an awakening – that Chinese companies are extremely good at building models in the same way the leading American companies are. Moonshot AI is solving many of the same problems that folks at OpenAI or Anthropic are solving. I’m confident there will be more distillation discussion, and pressure ：https://www.interconnects.ai/p/the-distillation-panic , but the evidence is now out that Chinese companies can do more than just fast following ：https://www.interconnects.ai/p/how-much-does-distillation-really .

Meeting some of the core Kimi team on my trip to China ：https://www.interconnects.ai/p/notes-from-inside-chinas-ai-labs , it was clear to me that they had incredible culture, some would say aura, and a freedom to express it – within the constraints of a GPU-limited environment. Where building models is so much of a scaling game, much of the ability to build a good model still comes down individual execution, motivation, and expression. Having visited them, this result is less surprising. Having visited many AI companies, very few have a culture that you can immediately pick up like this.

At the same time, China’s AI adoption trends started later than those in the U.S. So, while all the Chinese labs have way less compute than their counterparts in the U.S., more of it can certainly go to training. When I joked around about how much compute an average researcher at OpenAI could have – say a few thousand H100 equivalent machines – the researchers at Kimi were shocked. The org chart and approach to building the Kimi models surely reflect this, but it is difficult to tease out what this looks like without substantial proprietary information.

The state of affairs on peak model performance is roughly as follows ：https://x.com/finbarrtimbers/status/2077816750574539161 :

Moonshot AI – Kimi K3 (open weights*)

Zhipu (Z.ai) – GLM 5.2 (open weights)

Alibaba – Qwen 3.7 Max (3.8 announced, also to be open-weights, when writing)

It is astonishing to see DeepMind, and some of the other American giants this low. In many ways, the X AI team deserves more credit. A visual summary from Artificial Analysis is below:

This release and other recent events have caused a major change in direction for the most likely outcomes in the balance between open and closed models. I’ll unpack them individually.

In many ways, it feels like the start of a new era. An era with much more competition, but also a much higher need for coordination, as we rollout incredibly powerful technologies around the world.

Share ：https://www.interconnects.ai/p/kimi-k3-the-open-weights-escalation?utm_source=substack&utm_medium=email&utm_content=share&action=share

Many people started following China’s AI scene relatively recently, so they can reach the conclusion that releasing models openly is their core strategy. In fact, I think most labs have a core strategy far closer to Anthropic or OpenAI – build the best intelligence possible. Having followed and engaged with the Chinese labs for years now, the best explanation for their original turn to releasing their models openly is practicality. They needed to release the models openly to get adoption, attention, and feedback (especially in the high-value, Bay Area market) .

For a long time, there had been very limited policy in China ：https://www.chinatalk.media/p/chinas-new-ai-plan explaining the role of open-source AI, and what could be the “country-level strategy.” To my knowledge, no senior leaders had commented on open-source AI publicly. This changed this week too, as Xi Jinping gave a keynote address at the World AI Conference (WAIC), and very directly committed the future of China’s AI ecosystem to open-source and global diffusion ：https://mattsheehan.substack.com/p/xi-jinpings-big-ai-speech-annotated . This commitment to the status quo, the same week as the announcement of the strongest open-weight model to date, is a clear mark in the early history of modern AI.

This comes during a time period where many potential paths forward have been discussed for the Chinese AI industry – Will they stay open? Can they keep up with the American labs in scaling? Is there a growing revenue market in China? With these, the focus has been on China’s risk tolerance, the companies’ ability to monetize, and any closely related reason for a company to stop releasing their best models openly.

In tying Xi’s commitment in time to a very strong model, China has implicitly commented on its risk tolerance with respect to releasing open-weight models. For the time being, it is a read into the perceived risks of topics like strong cybersecurity capabilities (or bio-dangers) within the Chinese system.

The simplest explanation is that China’s government is definitely following potential risks from the models closely – likely with more technical scope than the US government’s vibe regulation – and would take action if it measured risk. The simple explanation is that they do not find current frontier models to have meaningful risk.

At the same time, China’s economic decision makers think having AI adoption is good, so they can make profits on the industry later – after growing distribution (as China has done for cars, solar, advanced manufacturing, and many areas in recent history).

These can seem somewhat shocking, in an American AI media landscape that has gone through months of hype and fearmongering over the Claude Mythos model. This surprise should be excellent grounding – the world does not have a unanimous agreement with the narratives about AI that we hear most in the U.S.

Many of the narrators guiding the discussion on AI have clear incentives to depress the perceived capabilities of the best, open AI models. Dean Ball – who is personally supportive of open models, but now works at OpenAI – had a widely commented on post ：https://x.com/deanwball/status/2078133895766114412 with some reflections on Kimi, where he said the following on open models. It is important to understand the statement, as it focuses the role of open models in the economic side of the AI buildout. Dean says: 1：#footnote-1

Open-weight models are inherently decelerationist, and I’m continually surprised to see the so-called “accelerationists” so excited about open-weight models.

Explaining why open-weight models are a form of decelerationism is important to understanding the coming world order. He is right.

Open models are decelerationist economically for the frontier labs, which will slow the net investment and capex rollout for AI. This is due to the fact that strong open-weight AI models massively reduce the margin potential for the closed labs. This has two effects. First, the AI labs have fewer profits to re-invest into future models. Second, the market sees the terminal value of these companies as being lower, so they will kneecap future fundraising rounds. These together will slow timelines to the most transformative AI models, but I do not see them as strong enough effects to stop OpenAI and Anthropic from being a few of the top valued companies in the world.

These, to me, are a net good for society. As open-weight models are accelerationist ：https://x.com/natolambert/status/2078164535262040448 for AI diffusion across the economy by having the entry price for intelligence at a certain level of performance be lower. Open models also encourage customization. The thing is that this type of diffusion is by its nature far slower than the frontier AI labs products, who sell tools used directly by developers. The potential for open models is for nearly every business to use them to craft domain-specific agents. This economic diffusion takes an extremely long time! I’ve described this as open-weight models being on a much slower starting, but potentially bigger exponential ：https://www.interconnects.ai/p/open-and-closed-models-are-on-different . The problem is, if closed models get too far ahead in raw capabilities, this ability to customize can be moot.

The combination of increased diffusion and decreased concentration of power in the AI labs I see to be very positive for the AI transition. It gives us more time to figure out the hard problems of new capabilities and lets more stakeholders impact the story – any one company is very likely to have issues with controlling the world’s most important technology safely.

It is, of course, important to me in this world for the best models to still be made by the U.S. companies, which will allow the US to control the trajectory of the technology and its values. I also expect this to be the case, as the U.S. has larger capital markets that are willing to invest in AI (and a growing share of profits), but it is not a given.

Kimi’s launch blog has some technical details that confirm the sort of improvements that are supplying the consistent model improvements we feel. To select one:

Kimi K3 is built on Kimi Delta Attention (KDA) and Attention Residuals (AttnRes), two architectural updates designed to improve how information flows across sequence length and model depth. We have also scaled up Mixture of Experts (MoE) sparsity, effectively activating 16 out of 896 experts when paired with a Stable LatentMoE framework. Together with refined training and data recipes, these structural changes yield an approximate 2.5× improvement in overall scaling efficiency compared to Kimi K2 , allowing the model to convert compute into intelligence more effectively.

Training efficiency really adds up. They will result in continued, incredible steps for the models.

As an aside, tracing the path of this particular innovation through the ecosystem is an interesting example. Kimi Delta Attention (KDA) was introduced in the Kimi Linear paper ：https://arxiv.org/abs/2510.26692 , which is similar to the Gated DeltaNet used for Olmo Hybrid ：https://arxiv.org/abs/2604.03444 (my last Olmo model while at Ai2). Qwen’s latest models switched to a related architecture and the recent Nemotron models also are hybrid (but still closer to Mamba than Gated DeltaNet). It’s awesome to see new architecture ideas like these, which were heavily progressed by academia, get so quickly translated into frontier-scale models. Gated Delta Networks ：https://arxiv.org/abs/2412.06464 were introduced in late 2024, building on ideas from Mamba. By mid 2026, they’re in frontier models.

I chose to focus on this example, partially because the Kimi team put a cool number to innovations between models, but primarily to give space to a broader discussion of China’s resource efficiency.

It is becoming clear that the Chinese labs are far more capital efficient. In a world where scaling laws dictate that intelligence is proportional to effective capital – which buys compute, data, & talent – that may be the greatest strength your AI industry could ever have. There are many possible explanations for why this is the case, such as Chinese researchers being paid less while being more effective at LLM research puzzles, but we will probably never get such specific reasons.

Since writing my notes ：https://www.interconnects.ai/p/notes-from-inside-chinas-ai-labs on China, I’m hearing more about an emerging data industry in China (far behind the billion dollar budgets of Anthropic for data) and that Chinese labs have access to meaningful training compute (by skirting export controls). Chinese companies do not have the same inference demand (until recently, as Moonshot AI had to pause new subscriptions ：https://x.com/Kimi_Moonshot/status/2078855608565207130 for access to their K3 model - while the API is still live), so much more of their compute could go to training. These areas impinge heavily on the truth of the ability of the labs, but we have very limited measurement into them.

The facts on the ground are that these Chinese labs have raised orders of magnitude less capital than any slice of the American AI ecosystem. The most direct comparisons are to OpenAI and Anthropic, who have slightly better public models. Others, such as Google and Meta have the largest cash flows in the history of business, and are behind on building models. As for American neolabs, the picture is even more competitive – Thinking Machines released their first model recently, Inkling ：https://thinkingmachines.ai/news/introducing-inkling/ , which is strong but not in the same class as Kimi K3.

These American companies with more resources could still catch up, but you need to strongly weigh the public measurements we have of model quality and not resort to hope – which often reflects a bias. If the Chinese labs do have a latent advantage, they could continue to utilize that to build even stronger models than all the competitors! Many outcomes are plausible and K3 should increase most people’s probability that China can outright lead in AI capabilities in the near future on the back of more efficient training efforts – even if it’s not your most likely predicted outcome.

A big contributor to the capital efficiency is likely in the approach, where American labs are spending meaningful energy in pushing the frontier in dramatic, big steps, and the Chinese labs are more focused on catching up — this catch-up is cheaper. Just as the student model can outperform the teacher in distillation generally (not limited to the adversarial distillation of the Chinese labs), an approach of “trying to catch up” rather than “invent the next paradigm” could lead to stronger models.

The weekend after the Kimi K3 release, while writing this and discussing the events broadly, Alibaba announced ：https://x.com/Alibaba_Qwen/status/2078759124914098291 that a 2.4 trillion parameter Qwen 3.8 model is coming soon with open-weights. Historically, Alibaba has kept their largest models as API-only offerings via their cloud business, so this is another big vibe shift opening the doors to the next chapter of the open model economy. Even if the model is behind Kimi K3 on benchmarks, it signifies that Chinese companies may not only be maintaining the status quo for their open model strategy, but leaning further into it.

If this Qwen 3.8 model releases soon, i.e. before the next Gemini model, it could push Google to the 8th position on the leaderboard of labs with the smartest models – a list that China has been climbing. There are other rumors of more strong Chinese models soon, with DeepSeek V4 expected to graduate out of it’s “preview” version.

Leave a comment ：https://www.interconnects.ai/p/kimi-k3-the-open-weights-escalation/comments

I think that if Claude Mythos was released as an open-weight model today, the negative outcomes would be relatively minor. This is a somewhat challenging opinion to hold, as we have very limited public cybersecurity evaluations and it is a complicated ecosystem (and because I trust many people at Anthropic). I still stand by it. The risks have been over-hyped.

The problem is that this will not always be the case for the strongest AI models. Far stronger models are coming — and with them increased risks — so it is an incredibly safe equilibrium for the best models to be accessed in a controlled, closed manner several months ahead of similar open-weight models.

Open-weight models which are very controllable by the user will always be coming — you cannot effectively ban digital products, especially from bad actors — as AI training has proven globally accessible longer than many analysts expected.

Still, as I write this , the government continues to flirt with more measures aimed at restricting open-weight model s ：https://www.interconnects.ai/p/6-months-to-live-for-open-models in the U.S. The latest is from Axios ：https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi :

Behind the scenes: The Commerce Department last year considered adding multiple Chinese AI labs to its “Entity List,” which would effectively cut off U.S. access without a license, a source close to the administration told Axios.

The National Security Agency and White House Office of the National Cyber Director also considered putting out an advisory on Chinese AI lab threats last year, practically discouraging U.S. companies from using their tech, the source said.

The White House considered implementing an executive order saying U.S. companies could only host Chinese models if they could guarantee security and take liability if it were breached, the source added.

Commerce last summer also circulated draft rules within the administration leveraging its authorities to secure domestic supply chains to target Chinese open-source models, another source close to the administration said.

This would leave the U.S. in a very asymmetric state where the best models in the U.S. have guardrails on cybersecurity tasks, but global actors have access to great Chinese open-weight models to probe our defenses. This is one of many examples where banning open-weight models is not only harms the free markets of AI but also makes the ecosystem less safe in the short-term. There are other very bad outcomes, such as slowing the diffusion of AI applications and AI research, as I discussed above.

These equilibriums are very hard to maintain, especially as AI tools accelerate progress in the models, but it is important to maintain this status quo between open and closed. Having a model that is truly alone at the frontier in capabilities — something like Mythos when it was announced — also be open-weight poses serious risks as we go into the unknown of capabilities. Models are going to progress very fast and it is increasingly hard to measure their total capabilities.

We are then stuck in a world where we are trying to thread the needle on open models. It’s reasonable to not want something so powerful to be diffused globally in an instant, but meanwhile the makers of the models are incentivized to hype their capabilities , and their competitors are incentivized to hype their risks. It all comes down to careful measurement and proactive hardening of society to risk vectors.

This careening train of policy debates, model releases, and raucous reactions is only going to continue from today. We’ve been on a train of rapid progress, where all the key ideas of how AI should play out are tested, since the release of Claude Opus 4.5 last December, which sent us down the agentic pathway. The key to making good decisions here is evaluation capabilities , independent of the companies with the largest financial stakes. One of many actions needed then, as we enter the AGI era of AI governance ：https://www.interconnects.ai/p/welcome-to-the-agi-era-of-ai-governance , is an Operation Warp Speed style approach of bootstrapping state capacity (and other independent actors) that can evaluate models accurately, and study emerging risks.

Open-weight models, by accelerating the diffusion of capabilities, are a massive escalation in the good and the potential bad of AI. For now, the bad side of frontier language models has been largely hypothetical, but that will not always remain the case.

Having open weight models be slightly behind the closed frontier is our natural buffer to mitigate the risks. The key point is that we must collectively act to mitigate potential harms as they appear, and whether open-weight models are 3 or 6 or 9 months behind, that is still a very short timeline. If we regulate open-weight models heavy-handedly, I suspect much of the world will be lulled into thinking we no longer need to act. All we would’ve done is slightly delayed the inevitable — open models will continue to cross all the key capability thresholds eventually and regardless of legality.

Understanding and benefiting from this open-closed dance must be a collective action from the AI community across all sectors of power and influence over the coming years.

Kimi K3 is a watershed moment because frontier open-weight models are now real. Many hypotheses will be tested on where risks of open-weight models truly land — I suspect it’ll be narrower than many expect, and many risks of AI will still be proliferated by closed and “safer” APIs. The evaluation of these risks will evolve in time with an acceleration of AI’s integration in our economy. We cannot get one without the other, and we will continue to get both.

With this, 2025 was when open models started to be taken more seriously — especially when China leaped ahead with such a clear lead ：https://atomproject.ai/ — as people realized that it would not be a unipolar world, with only American, closed AI labs determining the trajectory. 2026 is when those previously discussed, potential risks and accelerations due to truly frontier, open-weight models landed.

A large portion of the response was for the fourth bullet, which compared the inevitable outcome of open models to AI communism, which I think missed the mark. Specifically, the use of the word communism without explanation caused much of the blowback.