Preprint

Optimize Cheap, Deploy Strong: Cost-Aware Cross-Tier Transfer for Evolutionary Optimization

Tal Oved Roi Pony O. Naparstek U. Barzelay

Aug 2026 · 0 citations · 31 references

Computer Science

TL;DR

This work contributes a cost-controlled characterization of when cheap-tier search substitutes for target-tier search, and where it fails.

Abstract

Evolutionary optimization of LLM prompts and agentic programs (e.g., GEPA) is dominated by fitness evaluation: scoring each candidate runs an answering LLM over a validation set, so the evaluator's price tier dictates total search cost. We restructure that search by decoupling the three roles an LLM plays, running the high-volume answering role on the cheapest tier, reserving a strong model for the rare reflection/variation operator, then exploiting upward cross-tier transfer to deploy the cheaply evolved prompt on a stronger target. We contribute a cost-controlled characterization of when cheap-tier search substitutes for target-tier search, and where it fails. Across four tasks (HotpotQA, IFBench, LiveBench-Math, HoVer) and eleven models in four model families, the resulting prompt matches or exceeds same-tier optimization while placing over 96% of search tokens on the cheapest tier, at 5.6-14x lower search cost, rising to 25-54x where reasoning tiers emit long chains of thought on every fitness call.

arXiv Semantic Scholar

View source

Similar papers

Book Open access Sep 2026

Which LLM to Fine-Tune? Agent-Driven Model Selection at Scale

It is shown that model selection is a recommendation problem, and AgentRec, a multi-stage retrieval-and-ranking framework that progressively narrows hundreds of candidate models using increasingly expensive but more faithful evaluation signals, is introduced.

Chen Luo, Yu-Lin Liu, Yi Liu et al. · Proceedings of the 20th ACM... · 0 citations

Preprint Aug 2026

Janus: An Algorithm-Evaluator Co-Evolution Framework for LLM-Driven Discovery under Expensive Evaluation Budgets

Janus is introduced, a framework that uses LLMs to co-evolve target programs and executable proxy evaluators to address label scarcity and extend evaluator-guided LLM discovery from tasks with cheap, scalable feedback to scientific domains where trustworthy evaluation is scarce and expensive.

Xi-Meng Liu, Qianlong Wang, Ying-Ming Mao et al. · 0 citations

Preprint Aug 2026

Exploit More, Explore Smarter for Budget-Constrained Agentic Search

Budget-constrained agentic search arises when an LLM agent must refine candidates under a small evaluation budget, because validation is expensive, generation requires multiple model calls, or both. In this regime, standard MCTS allocates budget poorly: exploration bonuses dominate at low visit counts, unpromising sibl...

Haoyang Fang, Bernie Wang · 2 citations

Preprint Aug 2026

EcoAgent-Bench: Evaluating Economic Decision-Making in Budget-Constrained LLM Agents

EcoAgent-Bench is introduced, in which every task specifies priced actions and an explicit budget, and results show that completion under a budget and economical action selection are distinct properties.

Jie Wu, Ming Gong, Feixiang Cheng et al. · 1 citation

Preprint Aug 2026

SCOPE-Router: Cost-Aware Open-Set VLM Routing for Execution-Oriented Tasks

CRM+RCCR, an architecture-agnostic cost-aware objective that encodes cost preference into continuous relevance targets through per-pair independent scoring, eliminating multi-positive dilution while regularizing queries with similar routing preferences to be closer in the routing space.

Tao Yu, Yi-Fei Qu, Zhi-Qing Cui et al. · 1 citation

#machine learning Preprint Sep 2026

Code-to-Harness: Distilling Black-Box Optimizers from Self-Play

Can an agent learn a numerical search strategy through executable practice and then transfer that strategy as text? We study low-budget black-box optimization, where unaided language models remain well below strong classical optimizers. During development, an agent repeatedly writes and evaluates optimizer programs. It...

Yi Wu, Zheng Ren, Zhi-Yu Hu et al. · 0 citations