Getting better at coding doesn't make a model better at everything else

A majority of users in this sub use LLMs for coding/agentic tasks and I see why a lot of value is put into them but many try to say "Well coding has improved therefore it can just use tool calling and/or just look up what the user needs if there's a degradation for general knowledge/reasoning" and that's just not the case. Many LLM usecases can't just be fixed by an improvement to coding and agentic tasks. Creative writing, multilingual capabilities, offline constrained environments and many more usecases can't be easily fixed with just coding/tool-calling. I know Qwen3.8 is a TREMENDOUS win for local LLMS but I still hope that Qwen4 will improve generalist capabilities to be closer to frontier closed models in all other aspects too. Gemma is great but it wouldn't be ideal if it is the only major generalist local LLM going forward.