The language you're using for a Large Language Model (LLM) can have a huge effect on its cost and create an AI divide between English speakers and the rest of the world. A recent study shows that, due to the way services like OpenAI, measure and bill for server costs, English-language inputs and outputs are much cheaper than those in other languages with Simplified Chinese costing about twice as much, Spanishing
Beyond the kernel-side bits around Loongson Binary Translation coming in Linux 6.6, there's also been LoongArch SIMD-related work with enabling LSX and LASX usage in the kernel. For Linux 6.5 there was initial LSX and LASX SIMD
