Beijing-based startup releases the world's first open-source model in the 3-trillion-parameter range, adding fuel to US-China AI tensions
A Chinese AI startup just dropped the largest open-source AI model ever built. Moonshot AI, the Beijing-based company backed by Alibaba, introduced its Kimi K3 model on July 16, making it available through its API and hosted platforms. The full model weights will be downloadable by anyone on July 27 under a Modified MIT license.
The model packs 2.8 trillion parameters, making it the first publicly available model to crack the 3-trillion-parameter range.
What makes Kimi K3 different #
The model uses a mixture-of-experts (MoE) architecture combined with something Moonshot calls Kimi Delta Attention technology. Instead of firing every parameter for every query, the system activates only the most relevant subset of its massive parameter pool, keeping inference costs manageable despite the model’s enormous scale.
Kimi K3 supports 1 million tokens, which means it can process roughly the equivalent of several full-length novels in a single prompt. That’s designed for heavy-duty tasks like long-horizon coding, advanced knowledge work, and deep reasoning.
The Modified MIT license means developers can download, modify, and self-host the model freely.
The market response was immediate and overwhelming. Moonshot AI had to new subscriptions just days after launch because demand outstripped capacity.
The geopolitical backdrop #
This release arrives during a period of intensified US scrutiny over Chinese AI capabilities. The Biden and subsequent Trump administrations have layered export controls on advanced chips headed to China. Once the model weights go public on July 27, anyone with sufficient computing power can run, fine-tune, and deploy K3 regardless of jurisdiction.
What this means for investors #
If Kimi K3 genuinely delivers performance parity with top proprietary models from OpenAI and Anthropic, and does so as a free download, that complicates the revenue story for companies charging premium prices for API access. Moonshot AI’s Alibaba backing means the model will likely receive continued development, optimization, and support.
Regulatory responses from Washington could accelerate. If policymakers view open-source releases of frontier Chinese models as a threat vector, we could see new restrictions on deployment, hosting, or integration of foreign-developed AI systems within US infrastructure.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our