all-in-all, love this setup

it's a little slow for daily use. so i'm saving up for a RTX 4090 or farm of M4s

avante is great, btw

here's just a little more about current window context sharing

it took a little digging but Perplexity was right all along and i just pasted wrong

classic meat-brain problems

but it works!

look, i ain't no hero
Perplexity writes all my lua

and it knows *exactly* how to integrate neovim and avante with @lmstudio

let's fire up cURL

LM Studio opens a local endpoint for requests with four "OpenAI-like" endpoints:

- /v1models
- /v1/chat/completions
- /v1/completions
- /v1/embeddings

we're using /chat/completions for our neovim + avante setup

watch your memory

running these models locally takes a TON of system resources

you may need to close a couple hundred Chrome tabs

no lie…
Claude 3.5 Sonnet still trounces (even the open reasoning models) at summary RAG summarization

check out the result I get from the one-prompt YouTube description generator that Cursor created for me