Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k

Post by OlegKostrikin » Fri, Jul 17, 2026, 11:29 PM UTC

Post by mips_avatar » Mon, Jul 20, 2026, 6:02 AM UTC

gorkish wrote: post #428

mips_avatar wrote: post #221

Earlier quoted context omitted.

The shared prompts are all cached so it's a cache read which is like 10x cheaper than a regular prefill

I'm gonna jump in here and say that the way Anthropic's KV cache works (prefix matching) doesnt guarantee this at all if the subagents are all put into flight at close to the same time. There's nothing in the agent SDK that allows you to instance more than one agent simultaneously, and the docs don't say anything special about KV cache use by agents in this scenario. You could probably engineer your own harness or wo…