Linear token pricing breeds bloated agents — rationalized harnesses could run at ~300k tokens
JoshPurtell · x · 2026-11-07
In a debate with lujasper, JoshPurtell argues third-party agent harnesses aren't inherently wasteful — linear token pricing is to blame. With linear pricing, harness designers have no incentive to trim context, so agents burn 1M tokens pointlessly. A rationalized harness plus inference setup could get closer to 300k tokens, roughly Codex's level, which is efficient precisely because OpenAI owns both the model and the harness.
More from coding & agent
- PSA: AI agents love slapping broad :has() rules on CSS, causing huge style recalcs — jh3yy · 2026-11-07
- YC-backed Trepo debuts kitchen hardware that lets AI agents shop for groceries in the background — ycombinator · 2026-11-07
- Cognition ships Devin Review 2.0: 5× cheaper, 1.5× faster PR review — smtabatabaie · 2026-11-07
- Air moves its agentic development experience directly into JetBrains IDEs, free tiers included — therealdanvega · 2026-11-07
- PSA: keep CSS :has() rules tight — agents' broad selectors trigger huge style recalcs — jh3yy · 2026-11-07
- Solo founder credits AI chief-of-staff bots for record-breaking revenue month — joshgonsalves_ · 2026-11-07