The Handoff Tax: Continuing Non-Native Trajectories in LLM Agents

Hugging Face Daily Papers Papers

Summary

This paper studies the cost-quality trade-off when handing off tasks between low-cost and high-cost LLM models, introducing the 'handoff tax' concept where full trajectory transfer during escalation yields suboptimal quality recovery at a cost premium.

Coding agents perform long-running tasks spanning dozens of model calls, tool uses, and code edits. As these runs unfold, users face a practical cost-quality trade-off: escalating to a stronger model when a cheaper one struggles, or downshifting once the hard reasoning is complete. Each switch requires the receiver to continue a non-native trajectory produced by another model. We study how this handoff affects quality and cost, and how varying the trajectory information inherited by the receiver changes the outcome. Using pairs of low-cost, low-capability (LC) and high-cost, high-capability (HC) models from the Claude and GPT families, we vary handoff direction, timing, and interface, comparing full-trajectory transfer, compaction, and trajectory removal while preserving the repository state. Across both model families, full-trajectory escalation recovers less than half of the LC-to-HC quality gap while incurring a substantial cost premium. We term this cost-quality penalty the handoff tax. By contrast, downshift offers a favorable cost-quality point. Interestingly, the preferred interface also reverses with direction: reducing LC-model trajectory information improves escalation quality, whereas removing the HC-model trajectory reduces downshift quality.
Original Article
View Cached Full Text

Cached at: 08/27/26, 07:17 AM

Paper page - The Handoff Tax: Continuing Non-Native Trajectories in LLM Agents

Source: https://huggingface.co/papers/2608.24358

Abstract

Codingagentsperformlong-runningtasksspanningdozensofmodelcalls,tooluses,andcodeedits.Astheserunsunfold,usersfaceapracticalcost-qualitytrade-off:escalatingtoastrongermodelwhenacheaperonestruggles,ordownshiftingoncethehardreasoningiscomplete.Eachswitchrequiresthereceivertocontinueanon-nativetrajectoryproducedbyanothermodel.Westudyhowthishandoffaffectsqualityandcost,andhowvaryingthetrajectoryinformationinheritedbythereceiverchangestheoutcome.Usingpairsoflow-cost,low-capability(LC)andhigh-cost,high-capability(HC)modelsfromtheClaudeandGPTfamilies,wevaryhandoffdirection,timing,andinterface,comparingfull-trajectorytransfer,compaction,andtrajectoryremovalwhilepreservingtherepositorystate.Acrossbothmodelfamilies,full-trajectoryescalationrecoverslessthanhalfoftheLC-to-HCqualitygapwhileincurringasubstantialcostpremium.Wetermthiscost-qualitypenaltythehandofftax.Bycontrast,downshiftoffersafavorablecost-qualitypoint.Interestingly,thepreferredinterfacealsoreverseswithdirection:reducingLC-modeltrajectoryinformationimprovesescalationquality,whereasremovingtheHC-modeltrajectoryreducesdownshiftquality.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2608\.24358

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.24358 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.24358 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.24358 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

@rohanpaul_ai: New Stanford paper argues that, under equal reasoning budgets, one LLM usually solves multi-hop problems better than ma…

X AI KOLs Timeline

A new Stanford paper shows that under equal reasoning token budgets, single LLMs typically outperform multi-agent systems on multi-hop reasoning tasks, with gains from multi-agent setups often stemming from additional compute rather than architectural superiority. The paper uses the Data Processing Inequality to explain why information loss in handoffs harms multi-agent performance, and identifies context quality as the key factor where multi-agent systems can provide benefits.

Not All Tokens Are Equal: Inflation-Aware Routing for Agentic LLM Systems

arXiv cs.CL

This paper introduces InflationAgent, a routing system for agentic LLMs that measures token inflation, predicts task difficulty using CoT Branching Entropy, and optimizes model selection to maximize accuracy per cost, achieving higher accuracy with fewer tokens on benchmarks like GSM8K.