Latency and tooling support ? UX of cloud based LLM vs local is much better for the cloud option - not so much for dev tooling.

I tried using remote workstations - I am not a fan of lugging a beefy client machine to do my work - would much rather use something thats super light and power efficient.