20 toks per second is more than enough for agents.
You’re comparing cloud chat speed to agents and finding it lacking.
It’s not slow.
Agents work differently, and force you to work differently. You simply set a goal and let them work and come back later.
Their tool calls take up a lot of time in lag that has nothing to do with tok speed.
20 toks per second are more than enough.
When I teach ordinals and NFTs in my bitcoin workshop I use the example of Salvador Dali paying with checks and drawing on them. This is basically what an ordinal is. Written art on a Read more
My AI Hermes setup has expanded: Admin Hermes on Ubuntu server on miniPC so it’s always on and running crons. I use deepseek 4 flash to run things by opencode https://georgesaoulidis.com/opencode It connects to my Read more
0 Comments