Tokenflation: When "Hi" triggers 33 tool calls

6 pointsposted 9 hours ago
by jakozaur

1 Comments

cyanydeez

3 hours ago

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 ~30s