A new Chinese artificial intelligence model, Kimi K2 Thinking, has been released by the Beijing-based company Moonshot, claiming to outperform established models like OpenAI’s GPT-5 and Anthropic’s Claude Sonnet 4.5 on various benchmarks. The model, trained with around 1 trillion parameters, integrates long-horizon planning and adaptive reasoning, allowing it to effectively tackle complex academic and analytical problems. Notably, Kimi K2 is an open-source tool, which means it can be accessed and utilized for free, a significant contrast to the costly proprietary models that dominate the market. Moonshot asserts that developing Kimi K2 cost around $4.6 million, significantly less than the billions spent by major U.S. AI labs. If verified, this could shift the landscape for businesses considering AI tools, as they now have a free option that may surpass existing paid models.
Google’s Gemini now offers a feature that allows users to conduct “deep research” by accessing emails, documents, and chat logs to create comprehensive research reports. This functionality, highlighted as a key enhancement, enables users to initiate market analyses and competitor reports by integrating data from various Google Workspace products. The company announced that users can select from options including Google Search, Gmail, Drive, and Google Chat to inform their research queries. Currently available on desktop, this feature is set to roll out to mobile devices soon. This integration aims to streamline the research process, making it easier for users to compile and analyze information from multiple sources in one cohesive report.
Why do we care?
So, China just dropped an AI bomb. A company called Moonshot claims its Kimi K2 Thinking model outperforms GPT-5—and it’s open source. Trained for less than five million bucks. If that’s even close to true, it’s not just a tech milestone—it’s a strategic move, assuming the results are confirmed.
Imagine the CCP driving AI costs to zero in the West—same playbook as hardware manufacturing. Flood the world with free, capable models, and suddenly, billion-dollar Western labs look… inefficient.
Now, pair that with Google’s Gemini update—deep research across your Gmail, Drive, and Chat. It’s all about ecosystem control—closed and monetized versus open and potentially subsidized.
For IT service providers, this is your wake-up call: the AI race isn’t just about who’s smarter—it’s about who you can trust. Test the tools, but isolate your data. Build client advisory services around governance and compliance. Because as AI prices collapse, trust becomes your new premium service.
Across every story today, one theme stands out — AI’s value isn’t in the algorithms, it’s in who validates them.