One Config Change Said to Make Claude Code 3.1x Faster via Experimental Code Mode
damnGruz · x · 2026-11-15
A retweeted thread claims enabling an experimental code mode in Claude Code speeds up workflows by 3.1x. The steps given:
- Run claude --version; requires v2.1.323+, otherwise run claude update.
- Add "env": { "CLAUDECODEEXPERIMENTALCODEMODE": "1" } to /.claude/settings.json.
- Start claude in the terminal and choose to turn on code mode.
- Pick the model (the post says opus 5.5 and sonnet 5.5 also work).
- Give it a multi-step task, e.g. changing every file that calls one function, and press ctrl+o to see the generated code.
The post also links the author's article claiming the setup grants "super powers," without further technical detail. The model version names are unverified.
Related event: Claude Code's Experimental Code Mode Speeds Up Workflows(2 posts)→
More from coding & agent
- DeepMind engineer lets Gemini 4 Carbon burn unlimited tokens to auto-debug a Kaggle PR — thursdai_pod · 2026-11-15
- Ditching Claude Code ban anxiety: Factory.ai as an alternative coding agent — vista8 · 2026-11-15
- omarsar0: Without an 'eval factory' there's no 'software factory' — agent automation is worthless without evals — omarsar0 · 2026-11-15
- MaxxBench: tiny task tweaks spread coding-agent scores by 23 points — MFarajtabar · 2026-11-15
- Managing multiple Codex sessions is breaking down, Reddit user ventshelp — Excellent-Wheel7769 · 2026-11-15
- Anthropic says outdated prompting habits are slowing down Claude Opus 5.5 — The Decoder · 2026-11-15