news

rtk Raises Claude Code Costs at Low Effort: JetBrains Benchmark Debunks 60–90% Claim

A JetBrains benchmark (July 20) ran ‘rtk,’ a proxy marketed to cut Claude Code tokens 60–90%, across 425 billed trials. At low effort it made sessions a median 7.6% MORE expensive—while rtk’s own analytics logged 96.2M tokens ‘saved.’

Published 2026-07-21Source: Tech Times
Tech Times source artwork

Why it matters

Researcher Denis Shiryaev pinned Claude Code 2.1.201 and Sonnet 5, then measured the only tokens rtk actually compresses: uncached input barely moved (3.2%, p=0.23). More turns (+13.8%) and cache reads (+14.3%) turned a ‘saver’ into a small, consistent tax.

Tokenmaxxing read

Lesson for token-cutters: rtk rewrites ~a third of Bash calls, but Read/Grep bypass it entirely, capping savings near 3% of input. Replaying 83 transcripts predicted that ceiling for ~$0 before any paid run. Measure against real bills, not a counterfactual that never existed.

Source takeaway

Quality was a wash—71 of 80 low-effort tasks tied, sign test p=1.0—so rtk didn’t break code, it just billed more. It’s Shiryaev’s second debunk after the ‘caveman’ skill (9% saved, not 65%). Verify token-saver claims against instrumented baselines.

Topic links

Related projects

Tools that match this angle

#4In spirit
Agents

LangGraph

langchain-ai/langgraph

A framework for building resilient stateful agents with explicit graphs, persistence, human-in-the-loop flows, and controllable execution.

38.5K6.5KMIT
agentsstateworkflows
#1Direct
Routing

LiteLLM

BerriAI/litellm

An OpenAI-compatible gateway and SDK for calling many model providers with budgets, logging, load balancing, guardrails, and cost tracking.

55.1K10.2KSource-available
gatewaycost-trackingrouting
#2Direct
Observability

Langfuse

langfuse/langfuse

Open-source LLM engineering platform for observability, traces, metrics, evals, prompt management, datasets, and playground workflows.

32.2K3.4KSource-available
tracesevalscosts
Related feed

More source-linked context

Generated Tokenmaxxing editorial thumbnail for ‘I’m cancelling’: As Microsoft’s GitHub Copilot moves to token-based billing, developers fear rising AI costs - The Indian Express
newsTI
news

‘I’m cancelling’: As Microsoft’s GitHub Copilot moves to token-based billing, developers fear rising AI costs - The Indian Express

The Indian Express reports that Microsoft is moving GitHub Copilot from flat subscription pricing toward token-based billing, triggering developer backlash over the possibility of sharply higher monthly costs.

tokenmaxxingcoding-agentsagents
Read note
Fortune source artwork
long-formF
long-form

Microsoft reports are exposing AI's real cost problem: Using the tech is more expensive than paying human employees | Fortune

Fortune reports on a growing mismatch between “use AI everywhere” incentives and the reality that broad adoption can create surprisingly large bills—especially when agentic workflows multiply calls behind the scenes.

tokenmaxxingcoding-agentsagents
Read note
Generated Tokenmaxxing editorial thumbnail for First token counts reveal Opus 4.7 costs significantly more than 4.6 despite Anthropic's flat pricing - the-decoder.com
newsT
news

First token counts reveal Opus 4.7 costs significantly more than 4.6 despite Anthropic's flat pricing - the-decoder.com

Anthropic’s Claude Opus 4.7 keeps the same per-token pricing as 4.6, but real requests can cost more because the updated tokenizer can turn the same text into substantially more tokens.

tokenmaxxingcoding-agentsagents
Read note