GLM 4.6 in Claude Code to build a python app Took me around 3 mins to make this python app with a Streamlit web ui and llamacpp python bindings The app lets you run an inference via a simple chat interface plus configurable params 👇 (video 4x speed)

you can get any GLM coding plan using my link starts at $3/month

@streamlit Is there a reason we’d use GLM through CC interface vs let’s say, opencode?

@streamlit it’s not the same harness. if opencode works for you then it’s fine

@streamlit No I’m asking because I dont know, since you might have tried already. Not because I think one is better.

@streamlit i feel claude code is better and it’s free anyway. only diff is you can switch models in opencode

@streamlit I found it worked better on OpenCode

@streamlit I don’t get this. Is @cerebras using a different variation of the model

@streamlit @cerebras not sure what they did in cerebras but it seems different. some say it’s fast but not as good

@streamlit @cerebras Interesting Where are you subscribing to for the model? I’ll try that

@streamlit Currently glm4.6 doesn’t support input images. Is there any way to pass image so that i can pass ui mocks or sketches for reference?

@streamlit mcp

@streamlit if i use Vision MCP, first input will go to glm 4.5v then it goes to glm 4.6 or will i get the direct code from glm 4.5v itself?

@streamlit how do you calculate tokens used? can you give statusline script

@streamlit you dont need it when you’re using the glm coding plan

@streamlit have you tried minimax m2 in claude code?

@streamlit no i dont have an account. $10 is just high for me to just test it and I haven’t seen examples of it just people talking about it in X

@streamlit can they build a AI image generator which runs on local and you can give any huggingface model for text to image

@streamlit if you have a gpu why not

@streamlit Impressive speed! I might give that a try later.

@streamlit Honestly, building a functional Python app in minutes shows how AI is rewriting what “rapid development” even means.

@streamlit Sorry GLM4.6 seems to have lot of marketing folks. Virtually GLM is useless. All the examples presented are very basic coding examples

@streamlit 3 minutes is so cool

@streamlit 3 mins is wild, building speed is key

@streamlit GLM 4.6 plus Claude Code shows how quickly LLMs can bootstrap full Python apps with interactive UIs. Turning ideas into working prototypes in minutes!