Cozy Compute

For those of you wanting to get into local ai, it's still not too late. Start with a gpu first and get two 3090's for $2,500. Why 2? So you can learn to operate multigpu setups and get a feel for what different quants run like.

The reason Qwen 4.0 hasn't been released yet is the same reason OpenAi has held back their models. Security concerns and fear that their models can be used to hack into systems ie turning off a countries refrigerator systems and causing famine.

There are so many delays because they are trying to patch the models offensive cyber abilities and trickling them out helps them catch bugs.

Here it is. The future of Ai really does come down to Nvidia vs Anthropic. The essential question is what is a commodity (worth nothing) and what is valuable.

Nvidia wins: The gpu is valuable and Ai/software is free and open source.

Anthropic wins: Ai/Software is valuable and is closed source and through an api. Compute is cheap but mainly rented to people from data centers and nothing local.

This is why I think it was Anthropic that targeted Jensen. Both companies are trying to wipe out each other's business.

This is a satire account. But never tell anyone you have GPUs and you do local Ai. Nothing good will come of it and you could have a lot to lose.

I'm really liking what I'm seeing out of Inkling 2 Small from thinking machines. Here is a three.js comparison with opus 5.5 of a cat and dog playing together. First up is opus 5.5 which nailed it.