China creates the world’s largest open-weight AI model Kimi K3, gives tough competition to America’s advanced system – china moonshot unveils world largest open AI model performance like us model
AI startup Moonshot has launched the world’s largest open-weight AI model Kimi K3. This model is designed for advanced reasoning and long coding, which has given a tough competition to American AI models like OpenAI’s GPT-5.5 and Anthropic’s Claude Opus 4.8 in terms of performance.
China’s AI startup Moonshot launched a new AI model Kimi K3. This is a model with 2.8 trillion parameters, which the company has described as the world’s largest open-weight AI system. Its performance is said to be equal to the best Fable model of America’s big company Anthropic. This shows that China is continuously coming up with new and powerful models to keep itself ahead in the AI race. By doing this it is coming at par with America’s most advanced system.
Powerful model at low cost
Let us tell you that companies like Moonshot, Z.ai and MiniMax are launching more powerful models at a very low cost. This challenges the long-held belief that Chinese developers are months behind their American peers.
Model Specification
According to reuters report (REF.), Moonshot said that Kimi K3 is the first open-weight model. It is designed for advanced reasoning, long-term coding and knowledge work. This model has a 1 million-token context window, allowing it to process and store much more information in a single prompt than models of previous generations.
How is it different from ChatGPT or Gemini models?
In the open-source model, anyone can download the entire AI system, run it on their computer, and make changes to it.
In contrast, closed-source models like ChatGPT or Gemini cannot do this. You can only use them online and cannot make any changes in them.
Open-source models allow users to download, run, and customize its base system, whereas proprietary or closed-source models do not.
The company said that in terms of GPU kernel optimization, Kimi K3 performed on par with Fable 5 and outperformed Anthropic’s Opus 4.8, GPT 5.6 Sol and GPT 5.5. This refers to technologies that maximize the use of AI hardware and minimize latency.
About the authorMona DixitMona Dixit is an experienced journalist in the world of technology and gadgets. She is currently associated with Times Online as the Principal Digital Content Producer. Active in the field of journalism for the last 9 years, Mona has closely observed and covered the changing landscape of technology and consumer electronics. As a tech journalist, Mona specializes in reviews of new gadgets, launch events and changes happening in the world of AI. She not only keeps a keen eye on the latest smartphones, but also deeply examines audio devices and wearables. Mona also has a strong understanding of consumer electronics, AI innovation and the smartphone market. He is an expert in converting complex things of technology into simple life-hacks (Tips & Tricks) and accurately analyzing future technology. Mona believes that the real measure of technology is not just its ‘specifications’ but ‘user experience’. His real skill is to translate complex technical terminology into simple and colloquial language, so that even a common reader can take full advantage of the technology. For them the ultimate purpose of technology is to make life simpler and better. Mona’s real skill is to convey difficult technical information to the readers in easy and colloquial language. He believes that the real purpose of technology is to make life simpler. After Bachelor of Science, Mona has done post graduation in Journalism and Mass Communication from IMS Noida, Meerut University. Before joining Times Online, she had worked with leading technology websites – Techlusive and Newsbytes.… read more