I can't use DeepSeek without signing up. Someone tell me what's behind this thin wall separating us?

Is it even worth me signing up?

Idk, haven't noticed much diff from o1.

What I like about it is that I can now run o1-mini counterpart on ollama, on my iron

Data collection on behave of the inhumane chinese dictatorship?

Just download it. It's open source...

I do not think it's worth it. Reasoning is annoying and slower. It's better to prompt a bit more specific, do the reasoning yourself. This - if used for automation.

I know, I know, it will "learn" eventually but Jesus. Code generated on it's first try:

same with chatgpt lil bro

@fruitfcker what is that / what are you doing? Who will learn what?

@retoor what's your take on o3-mini?

@NoMad same as deepseek, meh. I don't like the reasoning so much. Do not really care about it. I am enthousiast about the gpt4o-mini that i use for all AI projects. I made a program that pipes everything you give it to gpt4o-mini. Use case? 'cat your-source-file.c | r "prefix all global variables with my_project_ --stdin". If no --stdin arg it's just a chatbot configured by personality description in a .rcontext.txt file. Super fun to make quick software with for reviewing / refactoring / branding / checking if something is spam / clean up CSS (did mess up today tho). It can also execute bash commands.

I can't use reasoning models for this kind of stuff. So i consider those things meh, impractical.

DeepSeek may have won some tests but ChatGPT still always contains more information.

@retoor I had "it" (DeepSeek r1) write a fun little Python script to come up with suggested lottery numbers and using binomial distribution to choose the specific set of odd and even numbers. There are 3 main draws in the Australian lottery: Thursdays is called "powerball" while the remaining 2 is just regular lottery with additional "supplementary numbers". The if-then block looks like it was someone who's just starting to learn Python programming.