← Back to articles

RTK reports token savings, but our cost benchmarks disagree

2026-09-11 | original

read original get O'Reilly "Programming Rust" book → more articles

Why This Matters

RTK has become a hugely popular tool (79k+ GitHub stars) on the promise that compressing terminal output makes AI coding agents dramatically cheaper, and viral claims of 60% token cuts have spread widely. But independent benchmarking — JetBrains' SkillsBench and this test across 1,740 Terminal-Bench 2.1 attempts costing over $1,500 — found roughly no net savings: 5% cheaper for one model, 5% more expensive for another. It's a useful reminder that reducing tokens read is not the same as reducing the bill, since fewer output tokens can mean more turns or lower quality.

Key Takeaways

RTK's own README concedes cutting up to 90% of bash output is not the same as cutting costs by 90%.
Controlled tests (Claude Code with Fable 5.0, OpenCode with DeepSeek V4 Pro) showed a 5% cost drop and a 5% cost rise respectively — effectively a wash.
Popularity and viral metrics are poor proxies for real efficiency gains; agent tooling claims need independent, cost-based benchmarking.

Worth a Look

O'Reilly "Programming Rust" book — RTK is a Rust-based CLI tool, and if reading about it makes you want to build your own fast terminal utilities, this O'Reilly guide is the standard reference for the language. It covers ownership, concurrency and systems-level programming in depth, so you can go from curious to shipping your own tooling.

See O'Reilly "Programming Rust" book on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.

RTK (Rust Token Killer) filters and compresses terminal output before the AI agent reads it. With over 79k GitHub stars today, RTK is one of the most popular tools to make AI coding cheaper.

One X post saying RTK could cut Claude Code tokens by up to 60% reached 313K views.

Yet JetBrains’s SkillsBench run found no savings. The README has a disclaimer:

RTK cuts up to 90% of the bash output your agent reads. […] it is not the same as cutting your bill by 90%.

So “less terminal output” is not the same as “cheaper AI coding”. It can help, be a no-op, or backfire (more turns or lower quality). In this post, we present our findings after several days and over $1,500 spent on tokens.

How RTK works

RTK can rewrite Git, test, package and file commands the agent runs through its shell tool ( Bash in Claude Code, bash in OpenCode). Each rewrite returns a terser version of the same output.

For example, RTK keeps file names, sizes and permissions ( 644 means rw-r—r— ), but drops the owner and date:

$ ls -la /app/warriors -rw-r--r-- 1 root root 824 Sep 13 2025 g2-clear.red -rw-r--r-- 1 root root 487 Sep 13 2025 paper.red $ rtk ls -la warriors/ 644 g2-clear.red 824B 644 paper.red 487B

Testing RTK on Terminal-Bench 2.1

... continue reading

Explore topics: rtk claude code terminal-bench jetbrains skillsbench opencode