Comment by baigy

Specifically: to explore your opensource options with compute limitations, ask the community at r/LocalLLaMA on reddit. That's where the current SOTA opensource text-to-text models live.