Thanks for helping improve Awesome LLM Token Reduction! This list collects high-quality techniques, tools, and research for reducing LLM token usage, with a focus on AI coding assistants.
- Fork this repository and create a branch.
- Add your entry to the most appropriate section in README.md.
- Open a pull request describing what you added and why it fits.
Each entry is a single bullet:
- [Name](https://canonical-url) - Concise description of what it does Rules:
- Use the canonical source (original repo or paper page), not a mirror or aggregator.
- Keep descriptions present tense, factual, and concise (target ~120-180 characters before the badge), with a trailing period.
- Place entries alphabetically within their section.
- One entry per line; no duplicate URLs anywhere in the list.
An entry must meet at least one strong signal and none of the rejects.
Strong signals (need ≥ 1):
- Actively maintained (commit or release within ~12 months), or
- A peer-reviewed or widely-cited paper, or
- Notable adoption (meaningful stars, used in production or research).
Reject if:
- The link is dead or paywalled with no public information.
- It is off-topic or only tangentially related to token reduction.
- It duplicates or mirrors an entry already present.
- It is a pure marketing page with no substance.
- Verify every link resolves (HTTP 2xx/3xx).
- Confirm the Table of Contents still matches the section headings.
- Keep formatting identical to the existing entries.