1. Compress the context before the model sees it
The largest share of spend is bulky, repetitive context: build logs, test output, JSON fixtures, and file content carried from turn to turn. Headroom sits between Codex and the API and compresses that content locally before forwarding it: about 50% fewer tokens on the noisy items it compresses, typically 25-50% across a session. The compression is reversible, so the model can pull back the original whenever it needs the detail. One install covers the ChatGPT app, the CLI, and the IDE extension, because all three read ~/.codex/config.toml.
2. Make terminal output quieter
Every command Codex runs sends its full output back to the model, and a failing build can add thousands of lines that every later turn carries. RTK rewrites common commands into token-lean equivalents, so Codex reads a summary of git status or a test run instead of the raw dump. Quieter test reporters and keeping generated files out of the working tree help the same way.
3. Stop re-reading the repository and documents
Exploring a codebase by grepping and opening files pulls in far more than the answer needs. Graphify builds a queryable graph of the project so Codex can look things up instead of reading files whole, and Engram keeps useful context across sessions, including through compaction, so each new session does not start from scratch. For PDFs and Office files, MarkItDown converts them to plain Markdown before Codex reads them.
4. Make the agent write less
Output tokens drain the same windows as input. Ponytail is a plugin that pushes Codex toward the smallest change that solves the problem, which means shorter diffs and less to review. Caveman shortens its prose replies. Both work on the response side, which the tools above leave alone.
5. Watch the windows instead of upgrading
When you hit a limit, OpenAI offers credits or a bigger plan. Before paying, check where the tokens went. The Codex usage limits guide explains the 5-hour and weekly windows, and why Codex is so expensive covers the patterns that drain them.
What to expect
Short, focused prompts will not get much cheaper. Noisy work will: debugging from logs, shell-heavy tasks, and codebase exploration are where these tools remove the most. Headroom alone typically removes 25-50% of a session's tokens, and under token metering that converts almost directly into more work per window; the other tools add to it on the sources they target. The benchmarks show savings by workload type. Judge them on a real task you already do rather than a toy prompt, and compare the token count before and after.
The Headroom app runs Headroom for you and turns on RTK, MarkItDown, Ponytail, and Caveman as one-click add-ons, so you do not have to install and update each tool yourself.