OpenAI Codex Dev: Skip Subagents — Good Compaction Means One Model Is Usually Enough
pvncher · x · 2026-10-01
pvncher from OpenAI's Codex team says don't use subagents for everything — not because they're bad, but because Codex models are strong on their own and compaction works so well that the bar for reaching for subagents is much higher than with Claude. Subagents remain useful for genuinely hard problems.
More from coding & agent
- New tool makes benchmarking across factory configs trivial, not just models — vikvang1 · 2026-10-01
- Magnitude inference engine hits #1 on HN, claims up to 2x faster local open-model runs than llama.cpp — nickbaumann_ · 2026-10-01
- marimo-lens lets you point at charts and steer your coding agent directly — S_Conradi · 2026-10-01
- Dev roasts Claude Code's Opus for forgetting UI parameters, makes No Country meme — firasd · 2026-10-01
- Solo fine-tune of Qwen3.8-27B-pi fixes effort ordering, saves 41% tokens — victormustar · 2026-10-01
- TinyFish browser agent kills the login wall: persistent sessions at $0.015 per step — SimplyAnnisa · 2026-10-01