Chinese startup Moonshot AI has released Kimi K3, a 2.8-trillion-parameter open-weight artificial intelligence model [1].
The launch represents a significant step in Beijing's effort to achieve AI dominance. By releasing the model's weights, Moonshot AI aims to challenge the proprietary lead held by U.S. firms and provide a massive open-source alternative for developers.
Moonshot AI unveiled the system in Beijing on Friday, July 12 [2]. The company said Kimi K3 is the largest open-source model ever released [3]. With 2.8 trillion parameters [1], the system is designed to compete with frontier models from American companies such as OpenAI and Anthropic.
Performance benchmarks suggest the model is closing the gap with U.S. rivals. Some reports indicate the system delivers performance approaching Anthropic's Fable model [4]. However, a Moonshot AI spokesperson said Kimi K3 still trails Anthropic's Claude Fable 5 and OpenAI's GPT 5.6 Sol on overall performance [5].
The release comes amid a broader national strategy to secure technological leadership. A Chinese government official said Beijing has made AI leadership a national priority, targeting global dominance [6].
Open-weight models allow external developers to see and modify the underlying architecture. This approach differs from the "closed" models used by many U.S. giants, where the internal workings remain secret. Moonshot AI is positioning Kimi K3 as a tool to accelerate AI adoption and innovation within China, and across the global open-source community.
“Kimi K3 is the largest open-source model ever released.”
The release of Kimi K3 signals a shift in the AI arms race toward massive-scale open-source systems. By providing a 2.8-trillion-parameter model, China is attempting to commoditize high-end AI capabilities, potentially undermining the competitive moat of closed-source U.S. providers while rapidly advancing its own domestic infrastructure.
