Originally, not all AI can speak Chinese well CodeX's weekly quota ran out again in the afternoon, and Antigravity's Gemini 3.1 Pro also took a hit. Helpless, I offer up a MiniMax that I really don't trust. At first, I just fixed CSS and found that the effect was okay; Then we began to delve into the backend logic of the program. Trust is gradually earned. This statement is absolutely true when it comes to MiniMax. In addition to outstanding coding ability, one of the elements that truly makes me feel "reliable" is: LLM can also speak Chinese well. Since the release of GPT 5, it seems that the entire LLM community has started to lose the ability to speak Chinese seriously, with a lot of "jargon" that makes viewers feel uncomfortable. I'm certainly not the only one roast about it. Some people say that this is all caused by contaminated Chinese corpus, while other LLMs are crazily distilling GPT 5 models behind the scenes, ultimately leading to a very tragic situation: Suddenly, all LLMs couldn't speak Chinese anymore. But obviously not everything. Today's test showed that the writing style of MiniMax is quite neat and pleasing to the eye. So, ultimately, where does the problem lie? A group of Chinese speaking AI experts in Silicon Valley, under heavy pressure, have tilted their resources towards learning and strengthening English language corpora, while completely investing in Chinese language corpora? Anyway, the boss doesn't understand Chinese, why don't we go dark under the light? Returning to the Chinese model, if the boss personally checks the clumsy writing style of the AI output, he must have already flipped over the table. Okay, just speak up and prepare to subscribe to MiniMax's Coding Plan for testing.