Skip to content
OpenTrain AIFor AI Companies
← Back to explorer

Token Reduction Is Not Cost Reduction

Sarel Weinberger, Amir Hozez · Jul 13, 2026 · Citations: 0

Automatic Metrics Coding Long Horizon Medicine
Open arXiv Find Implementation RSS feed Shortlist (0)

How to use this page

Moderate trust

Use this for comparison and orientation, not as your only source.

Best use

Background context only

What to verify

Validate the evaluation procedure and quality controls in the full paper before operational use.

Evidence quality

Moderate

Derived from extracted protocol signals and abstract evidence.

Abstract

Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost. We evaluate three token-reduction approaches against an unmodified Claude Code baseline across controlled coding tasks, measuring provider-billed cost, task success, cache traffic, and agent behavior. The largest compression setup reduced delivered tool-output tokens by 38.4% but increased billed cost by 6.8%, while lighter compression produced only small and statistically uncertain savings. Across tasks, token reduction was weakly correlated with cost reduction (Pearson r = 0.15). Cost decomposition shows that prompt-cache creation and reads dominate the measured input-side cost, leaving only a limited fraction of total spend directly addressable by tool-output compression. We also find that compression can alter agent trajectories through additional retrieval, diagnosis, testing, and turns, offsetting local token savings. On a SWE-bench Go subset, aggressive compression also reduced successful patch application. These results show that token reduction is not a reliable proxy for cost reduction in tool-heavy coding agents. Effective optimization should therefore be evaluated at the level of cost per successful task, including cache behavior, trajectory changes, and correctness rather than token counts alone.

Should You Rely On This Paper?

This paper is adjacent to HFEPX scope and is best used for background context, not as a primary protocol reference.

Best use

Background context only

Use if you need

A benchmark-and-metrics comparison anchor.

Main weakness

No major weakness surfaced.

Trust level

Moderate

Usefulness score

25/100 • Low

Treat as adjacent context, not a core eval-method reference.

Human Feedback Signal

Not explicit in abstract metadata

Evaluation Signal

Detected

Usefulness for eval research

Adjacent candidate

Extraction confidence 55%

If you are doing eval pipeline work, start here:

Human Eval Hub LLM-as-Judge Hub Pairwise Preference Hub Tool-Use Eval Hub

What We Could Verify

These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.

Human Feedback Types

missing

None explicit

No explicit feedback protocol extracted.

"Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost."

Evaluation Modes

strong

Automatic Metrics

Includes extracted eval setup.

"Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost."

Quality Controls

missing

Not reported

No explicit QC controls found.

"Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost."

Benchmarks / Datasets

strong

SWE Bench

Useful for quick benchmark comparison.

"On a SWE-bench Go subset, aggressive compression also reduced successful patch application."

Reported Metrics

strong

Task success, Inference cost

Useful for evaluation criteria comparison.

"Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost."

Human Feedback Details

  • Uses human feedback: No
  • Feedback types: None
  • Rater population: Not reported
  • Unit of annotation: Trajectory
  • Expertise required: Medicine, Coding

Evaluation Details

  • Evaluation modes: Automatic Metrics
  • Agentic eval: Long Horizon
  • Quality controls: Not reported
  • Evidence quality: Moderate
  • Use this page as: Background context only

Protocol And Measurement Signals

Benchmarks / Datasets

SWE-bench

Reported Metrics

task successinference cost

Research Brief

Metadata summary

Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost.

Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.

Key Takeaways

  • Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost.
  • We evaluate three token-reduction approaches against an unmodified Claude Code baseline across controlled coding tasks, measuring provider-billed cost, task success, cache traffic, and agent behavior.
  • The largest compression setup reduced delivered tool-output tokens by 38.4% but increased billed cost by 6.8%, while lighter compression produced only small and statistically uncertain savings.

Researcher Actions

  • Compare this paper against others mentioning SWE-bench.
  • Check the full text for explicit evaluation design choices (raters, protocol, and metrics).
  • Use related-paper links to find stronger protocol-specific references.

Caveats

  • Generated from abstract + metadata only; no PDF parsing.
  • Signals below are heuristic and may miss details reported outside the abstract.

Recommended Queries

SWE-bench

Research Summary

Contribution Summary

  • Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost.
  • We evaluate three token-reduction approaches against an unmodified Claude Code baseline across controlled coding tasks, measuring provider-billed cost, task success, cache traffic, and agent behavior.
  • We also find that compression can alter agent trajectories through additional retrieval, diagnosis, testing, and turns, offsetting local token savings.

Why It Matters For Eval

  • Token-reduction tools for coding agents are often evaluated by the number of tokens they remove, but token count alone does not determine end-to-end inference cost.
  • We evaluate three token-reduction approaches against an unmodified Claude Code baseline across controlled coding tasks, measuring provider-billed cost, task success, cache traffic, and agent behavior.

Researcher Checklist

  • Gap: Human feedback protocol is explicit

    No explicit human feedback protocol detected.

  • Pass: Evaluation mode is explicit

    Detected: Automatic Metrics

  • Gap: Quality control reporting appears

    No calibration/adjudication/IAA control explicitly detected.

  • Pass: Benchmark or dataset anchors are present

    Detected: SWE-bench

  • Pass: Metric reporting is present

    Detected: task success, inference cost

Related Papers

Papers are ranked by protocol overlap, extraction signal alignment, and semantic proximity.