It’s a mistake to conflate “wants to spend money on the most ethical option available” with “ think the most ethical option available is perfect”

Why wouldn’t you move your dollars to someplace incrementally better?

You make it sound as if "the most ethical option available" is.. actually ethical?

Their statement doesn't make it sound they are incrementally better, they are trying to bend over backwards to keep working for war.

I am not greatly relieved by this post of Anthropic's. That said, they seem to have lines and are willing to stand by them; I don't see where OpenAI has done that. So, for now and from my point of view, the point goes to Anthropic.

Moving my subscription is not terribly consequential, but since the products are so similar and easy to substitute with one another for my uses, it seems best to participate in what in aggregate is a signal that is being noticed and commented on and interpreted to mean that a significant number of people who buy AI access do care about this.

There are so many inference providers not working for Department of War. Even Alibaba and sure China has lots of issues but they are not bombing anyone now if that's your first priority. Or else, smaller US / European / Asian companies with pure civilian focus. SOTA open weights models they serve are perfectly suitable for coding and chat. I run a local Qwen3.5-122B-A10B-NVFP4 instance and it writes entire Android apps from scratch and that's a midsized model.

Sorry for the off-topic but what hardware are you running Qwen3.5-122B-A10B-NVFP4 on? Is it physically local or just self-administered? Thanks in advance.

I have an NVIDIA Thor Dev Kit, a somewhat less known cousin of DGX Spark.

I'm not sure there's really any good large model providers

Can you give a list of high quality alternatives? Morally speaking i would put China on par with the US if not worse (due to their ongoing Uyghur genocide). I will check out Qwen3 but would be interested in others.