[ukuina]

Low-latency multi-modal LLMs with constrained outputs can outperform most specialized prev-gen networks, so this may actually work out independent of achieving AGI!