Claude Code users report 3.8x token efficiency drop in a week, suspect secret orchestration change
Wise_Carpenter_3132 · reddit · 2026-10-27
A Reddit user's token analysis of Claude Code shows efficiency dropped 3.8x week-over-week: the same weekly limit now yields 31K output per 1% vs 119K last week, and this week's output would have cost only 17% of the limit at last week's rate. Three anomalies: subagents re-read more from the parent, take 2x steps to finish, and more subagents spawn per task. He suspects an unannounced agent orchestration change (or a model quality drop), and suggests disabling subagents in the main session and restarting at 300-400K context, hoping Anthropic offers a weekly reset as compensation.
More from coding & agent
- Viral screenshot shows an agent spawning 32 parallel workspaces in ~30 seconds — cephaloform · 2026-10-28
- Grok Bot 0.78.1 hides unreleased per-bot logins with manager-only team sharing — faraz0x · 2026-10-28
- Uber Built a Legal Redlining Agent for Its Lawyers—Here's What It Learned — AxSaucedo · 2026-10-28
- GraphRAG's hub-node trap: same two-hop Cypher costs 71 DB hits normally, 60,013 on a hub — JeremyCMorgan · 2026-10-28
- Dev claims his memory system beats Claude Code's, which he says Anthropic took from him — RileyRalmuto · 2026-10-28
- Grok Imagine team opens up their AI short-film workflow with a live prompting guide — tetsuoai · 2026-10-28