Moonshot AI released Kimi K3, a 2.8 trillion parameter model that the Chinese startup says closes the performance gap with leading US AI systems.

The model outperformed Claude Opus 4.8 and GPT 5.5 on coding and general agent benchmarks, though it still trails Anthropic's Claude Fable 5 and OpenAI's GPT 5.6 Sol on overall performance, according to the company.

K3 represents China's largest AI model to date. Bank of America analysts noted that despite "persistent hardware/compute capacity constraints in China", the model demonstrates that pre-training scaling paired with architectural innovation can still deliver significant gains for Chinese AI systems.

Market reaction mirrors DeepSeek moment

Patrick Moorhead, CEO of Moor Insights and Strategy, called the market reaction "an over-reaction shockingly similar to the DeepSeek panic". He argued that despite technological advances, "we are far away from super-intelligence".

The release highlights a broader industry shift from focusing solely on model size to building complete applications. Perplexity CEO Aravind Srinivas told CNBC that startups now prioritise finding the best methodologies for using AI models in their apps rather than relying on one massive system.

"The model alone is no longer the product," Srinivas said. "It is the harness, the orchestration system that puts the model inside a very capable harness and pairs the model with a lot of tools."

Lu Zhang from Fusion Fund noted that most developers using these models come "from the startup ecosystem, less from the large corporate side". These developers frequently swap models when more powerful or cost-efficient versions become available.

Open-weight competition intensifies

The Kimi K3 launch comes as Chinese AI models gain traction among Western companies due to competitive performance and lower costs compared to advanced American offerings.

US lawmakers are considering measures to curb adoption of Chinese AI models by domestic companies. The debate reflects broader US-China competition in AI development, though Zhang pointed out that several US companies are also releasing open-weight models.

Simon Koser from AI startup Tzafon called K3 "legitimately impressive" in coding performance, noting that cost has become a major consideration for AI labs seeking returns on investment.