Interesting. Always glad to see more open weight models.

I do appreciate that they openly acknowledge the areas where they followed DeepSeek's research. I wouldn't consider that a given for a US company.

Anyone tried these as a coding model yet?