The cheapest model optimization this month is a cli proxy, not a model. rtk, an open-source rust…
the cheapest model optimization this month is a cli proxy, not a model. rtk, an open-source rust binary with zero dependencies, cuts llm token use up to 90% on common dev commands by filtering output before it reaches the context window, with claude code and cursor support built in.
deleting tokens beats begging for a discount.
Context
The rtk README says it cuts up to 90% of the bash output your agent reads, and states that this is what RTK measures and is not the same as cutting your bill by 90%. It gives install commands for Claude Code, Cursor and OpenCode, and notes the hook runs only on Bash tool calls, so Claude Code's built-in Read, Grep and Glob bypass it. A JetBrains A/B test of 20 July 2026, run by JetBrains and not the rtk authors, measured rtk at 7.6% more expensive per task at low reasoning effort over 80 clean pairs and about the same at high effort, with quality unchanged, on Claude Code 2.1.201 with claude-sonnet-5. JetBrains also reports rtk's own savings counter overstates savings by counting the full raw output as the counterfactual.
The up to 90% is a cut in bash output per command type, not total tokens or cost. The independent test found no saving on that harness, model and effort, and does not generalize to other setups; it is not merged with rtk's own figures. Savings depend on how much work flows through Bash. Zero dependencies was not verified from the text read. Cheapest tokens being the ones never sent is the author's line.
Related work
- Earlier note on rtk ↗Same tool.
Watch next
- Independent tests on other harnesses.
Sources
- rtk README (GitHub)github.com
- RTK token savings test (JetBrains, 20 Jul 2026)blog.jetbrains.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 20 September 2026 at 01:49 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →