Kimi, GLM and MiniMax, the Chinese alternatives I use
What they do well, what they cost, and what data I give them and what I don't.
DeepSeek is a Chinese artificial intelligence company and also the name of its assistant and its language models, which compete with those from OpenAI or…
DeepSeek is a Chinese artificial intelligence company and also the name of its assistant and its language models, which compete with those from OpenAI or Anthropic. You can use it as a chat, just like ChatGPT, or through its API (the entry point for other programs to use the model), and some of its models are published openly for download.
What's confusing is that DeepSeek isn't just one thing. There's the chat app, there are the models, with versions that keep changing, and since 2026 there's DeepSeek Harness, an open-source tool for working with agents. Another frequent question is privacy: when you use its chat or its API, what you type goes to its servers, so it's worth reading its terms before putting in sensitive information. With the open models you run on your own machine things are different, although the biggest ones need a powerful computer.
DeepSeek is one of the AIs I work with, along with Claude, OpenAI, GLM, Kimi and MiniMax. In "Can you work professionally with free AI tools?" it's part of the review of what's free; in "Ten days working with DeepSeek Harness" I explain what it was like bringing its harness into my real work; and in "Kimi, GLM and MiniMax: the Chinese alternatives I use" I talk about the other Chinese AIs I use.
What they do well, what they cost, and what data I give them and what I don't.
What happened when I put it into my real work, compared with what I already use.
ChatGPT, Claude, Gemini, Copilot, DeepSeek, GLM, Kimi and MiniMax, and where the limit lies.
One email a week with what was verified, what was refuted and what is still on the bench. No news roundups.