To me, this is exactly the kind of evolution needed.

Brute-forcing a problem ("bigger == better") is the retarded approach to solutions but because the big companies are boasting their costs like its a benefit to society and people are jumping on that bandwagon, somehow a very large part of society really wants Elon to get another 50B in funding to waste on feeding X posts as truth to an autocorrect algo. You know what? For 50B, we can also feed in Reddit and 4chan. AI will make the world a better place with all that knowledge. lol.

Qwen is open weights. It is meant to be built upon, and so are Deepseek and llama (but there seem to be some issues with llama4, Zuck's probably off the open source path.) Not using that is a shame. I think that these open models can definitely use fine-tuning, and in open source, you just do it if you want it.

People using their big brains and doing smart approaches to problem solving in the space is just what we need. So that we can move away from building massive datacenters just to transform text to math in massive batches that we then throw away. I hope for more of this.