# Harnessing LLMs as Agents: What Does It Cost?

arXiv: https://arxiv.org/abs/2610.02488 · HTML: https://mlsanity.com/paper/2610.02488/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Agent memory systems, one of 16 directions in LLM agents and tool use.
- Builds on 29 of its 52 references that are in the map.
- First posted on arXiv 1 Oct 2026.

## Placed in

- LLM agents and tool use / [Agent memory systems](https://mlsanity.com/agents/agent-memory-systems/)

## Builds on (most cited first, 29 of 29)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [MemGPT: Towards LLMs as Operating Systems](https://mlsanity.com/paper/2310.08560/) (2023)
- [FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance](https://mlsanity.com/paper/2305.05176/) (2023)
- [Theoretical Limitations of Self-Attention in Neural Sequence Models](https://mlsanity.com/paper/1906.06755/) (2019)
- [Chain of Thought Empowers Transformers to Solve Inherently Serial Problems](https://mlsanity.com/paper/2402.12875/) (2024)
- [AI Agents That Matter](https://mlsanity.com/paper/2407.01502/) (2024)
- [Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering](https://mlsanity.com/paper/2604.08224/) (2026)
- [Memory Augmented Large Language Models are Computationally Universal](https://mlsanity.com/paper/2301.04589/) (2023)
- [Tracking Objects as Pixel-wise Distributions](https://mlsanity.com/paper/2207.05518/) (2022)
- [PENCIL: Long Thoughts with Short Memory](https://mlsanity.com/paper/2503.14337/) (2025)
- [Not All LLM Reasoners Are Created Equal](https://mlsanity.com/paper/2410.01748/) (2024)
- [Deliberate Reasoning in Language Models as Structure-Aware Planning with an Accurate World Model](https://mlsanity.com/paper/2410.03136/) (2024)
- [ProTo: Program-Guided Transformer for Program-Guided Tasks](https://mlsanity.com/paper/2110.00804/) (2021)
- [On the Query Complexity of Verifier-Assisted Language Generation](https://mlsanity.com/paper/2502.12123/) (2025)
- [Adaptive Information Control for Search-Augmented LLM Reasoning](https://mlsanity.com/paper/2602.01672/) (2026)
- [Grounding and Enhancing Grid-based Models for Neural Fields](https://mlsanity.com/paper/2403.20002/) (2024)
- [Augmenting Policy Learning with Routines Discovered from a Single Demonstration](https://mlsanity.com/paper/2012.12469/) (2020)
- [Laplacian Multi-scale Flow Matching for Generative Modeling](https://mlsanity.com/paper/2602.19461/) (2026)
- [CETCAM: Camera-Controllable Video Generation via Consistent and Extensible Tokenization](https://mlsanity.com/paper/2512.19020/) (2025)
- [DIVE: Unlocking Self-Improvement in Frozen Language Models Through Diversity-Driven Skill Evolution](https://mlsanity.com/paper/2608.12486/) (2026)
- [Perturbation Sensitivity of Maximum-Likelihood Pairwise Ranking in Computational Decision Systems](https://mlsanity.com/paper/2604.17805/) (2026)
- [Beyond Helpfulness: A Teaching-over-Solving Diagnostic for Measuring Educational Impact in LLM Tutors](https://mlsanity.com/paper/2606.16206/) (2026)
- [On the Combination of Silent Error Detection and Checkpointing](https://mlsanity.com/paper/1310.8486/) (2013)
- [Computability of Agentic Systems](https://mlsanity.com/paper/2602.13222/) (2026)
- [Reliable Financial Named Entity Recognition Under Domain Shift: Confidence Estimation and Selective Prediction](https://mlsanity.com/paper/2608.19558/) (2026)
- [Toward Workflow-Aware Benchmarking for Healthcare NLP Agents](https://mlsanity.com/paper/2609.00296/) (2026)

## Abstract (arXiv)

Language-model agents increasingly rely on harnesses that manage bounded context, persistent memory, tools, verification, and repeated execution, yet existing notions of model capability do not quantify the computational resources these mechanisms consume. We introduce the Language Model Agent Machine (LAM), a resource-bounded abstraction that fixes the underlying semantic model while explicitly charging harness-level resources. We establish four classes of results. Communication: LAM execution is instancewise equivalent to red--blue pebbling under simultaneous call--transfer budgets, transferring classical I/O lower bounds to context--memory traffic. Access: memory interfaces induce asymptotic separations, including a Theta(n) gap between random and non-speculative sequential access on pointer chasing. Recomputation: bit-reversal DAGs require Theta(n2/(C+S)+n) model calls with context capacity C and persistent-memory capacity S, quantifying when stored intermediate state avoids repeated semantic computation. Reliability: we derive tight stage-local sampling bounds, exact imperfect-verification costs, and a Young--Daly-type checkpoint law with a closed-form optimal verification interval. Controlled and held-out experiments on GPT-6 Astra test communication and reliability predictions, including checkpoint optima, policy selection under programmatic checking, and tradeoffs among call granularity, logical input traffic, and reliability on chained MATH tasks. Together, these results provide a resource theory for the computational cost of language-model agent harnesses.