"I'd like a car that goes 300mph and gets 100mpg while doing it. I'm aware of a car that gets 100mpg but it is extremely slow."

You are describing fundamental tradeoffs. Getting more performance relative to model size and training token amount is what all of the labs are solving.

▲

docheinestages 6 hours ago | parent [-]

Labs are focusing on creating models, small or large, that perform well on various benchmarks, including general knowledge, domain-specific expertise, and agentic capabilities.

Asking for such a model while wanting to be small and fast would align with what you're describing, which I believe is different from what I'm pointing to.

The model I'm describing sacrifices domain knowledge and expertise for agentic reasoning and tool-calling capabilities at a reasonable speed.

Think of Cactus Compute's Needle [1].

[1] https://cactuscompute.com/blog/needle

	▲	lossolo 4 minutes ago \| parent [-]
		You can't "sacrifice domain knowledge" because it's part of latent space reasoning. Connections are between concepts, and if you want to remove these concepts, you also remove connections, which means removing reasoning capacity of these models. Some reasoning patterns get established on reasoning in one domain and, based on similarity, get transferred to another domain inside the weights. These are not brains, they don't work like brains, and it seems you think about them as brains.