7/26/2026 at 1:33:32 PM
i can't believe people are waiting 5 minutes for a hi call on a cloud model. I assume most of that is provisioning; local model from cold start only takes ~30sby cyanydeez
7/26/2026 at 7:15:16 AM
Tokenflation: When "Hi" triggers 33 tool calls
https://quesma.com/blog/tokenflation-when-hi-triggers-33-tool-calls/by jakozaur