Qwen 3.7 is not open source; previous Qwen versions would have open source releases, but Qwen 3.7 plus does not. The second best Chinese model, Minimax M3, is testing the waters by taking longer and longer between “model release” and open sourcing it. This time, they spent 2 weeks after release before open sourcing it. There’s also a lot of rumors of GLM and Deepseek not open sourcing future models.

It’s pretty obvious that you cannot take Chinese models as open source for granted, they’ll be closed source soon.

This reads to me as "user installed exe file can upload your data to a server". Um, yes, that's the point?

This seems like this generation's equivalent of "don't open Linkin-Park.mp3.exe from limewire"

The tokens/sec of the model is basically directly proportional of the memory bandwidth of the hardware it runs on. So either OpenAI has to gimp model performance for its entire life, or somehow magically speed it up 4x on the first day.

I am willing to bet large amounts of money that OpenAI would never release a model served as fully BF16 in the year of our lord 2026. That would be insane operationally. They're almost certainly doing QAT to FP4 for FFN, and a similar or slightly larger quant for attention tensors.

More and more people are just focused on making a quick buck.

I'm getting a feeling that these people would gladly rip off a lemonade stand, and then defend themselves by saying the lemonade stand deserves it.

Hell, Dunbar's Number is 150 people, and you expect to have 50 directs? That's literally 1/3 of your 150 being occupied by directs. It seems clearly infeasible the more you think about it.

It seems quite counterproductive to assume such a system would scale to everyone else, or that everyone else could possibly implement this. This is cowboy levels of human resource management, not careful engineering.

[1] https://en.wikipedia.org/wiki/Seeing_pink_elephants

Therefore, you must not use humans for any important work.

It's pretty clear for me to argue that those things are NOT intuitive at all, and not easy to recognize. That's not changing the goalposts at all. Would the median american voter understand Poincaré's contributions to algebraic topology? Obviously not. Things that are easy for people to recognize: "touching a hot stove burns you". Things that are not easy for people to recognize: Poincaré's contributions to algebraic topology.

Honestly, your argument falls apart the moment you think about it critically. If it was so easy to recognize bias, then wouldn't all the people in the species already recognized it and voted to shape our legal system to handle any such bias, so it wouldn't be an issue right now? Clearly, that's not the case (we're still dealing with such issues), and understanding such biases is obviously an issue for people in the general public.

If you go to a tribe in the middle of the rainforest, would they be able to explain those concepts? Of course not.

Plus, I already gave an example of a species wide bias at the end of the comment- phone addiction for kids. I'm clearly not saying it's impossible for a human to spot a bias, but rather... how many 5 year old kids recognize that phone addiction is a bad thing?

[Citation Needed]

Just because if you have a species-wide bias, people within the species would not easily recognize it. You can't claim with a straight face that "we're really not that vulnerable to such things".

For example, I think it's pretty clear that all humans are vulnerable to phone addiction, especially kids.

I got Samba 4 working on Apple Time Capsules: https://github.com/jamesyc/TimeCapsuleSMB

If you have a legacy Time Capsule you'd rather not e-waste, you can try this out. Note that this is very much beta quality software, so don't expect it to work on all configurations.

This runs Samba 4 on the Apple Time Capsule.