It's Not the Capability: Harness Sensitivity Is Non-Monotone Across LLM Agent Tiers

• URL: http://arxiv.org/abs/2605.26731v1
• Date: Tue, 26 May 2026 09:08:41 GMT
• Title: It's Not the Capability: Harness Sensitivity Is Non-Monotone Across LLM Agent Tiers
• Abstract summary: A prevalent assumption in LLM agent deployment holds that more structured harnesses universally improve reliability.<n>We test this hypothesis through a controlled 432-run experiment crossing six models across four capability tiers.<n>We introduce a six-label failure taxonomy showing that format_violation dominates capable-model failures while wrong_file dominates low-capability failures.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: A prevalent assumption in LLM agent deployment holds that more structured harnesses universally improve reliability, and that higher-capability models need proportionally less structural guidance -- together implying a monotone inverse relationship between model capability tier and optimal harness complexity. We test this hypothesis through a controlled 432-run experiment crossing six models across four capability tiers with three harness conditions (light, balanced, strict) on HEAT-24, a 24-task synthetic benchmark with git-based workspace verification. Our results refute the monotone inverse relationship on two fronts. First, for the frontier chat model evaluated (Gemini 2.5 Flash), increased harness verbosity lowers VTSR by 29-38 percentage points -- a harness-complexity paradox. Second, for the frontier reasoning model evaluated (Qwen3.5-122B, extended thinking enabled), strict harness achieves the highest VTSR (91.7%) and the lowest latency, the opposite of the prediction. Within the constrained tier, a 2B model (Gemma4:e2B) matches strong-open-tier stability at 91.7% across all harnesses. Because each tier is represented by a single model in this study, these results should be interpreted as model-specific observations; harness sensitivity appears non-monotone across the models evaluated, and depends critically on model type (chat vs. reasoning). We introduce a six-label failure taxonomy showing that format_violation dominates capable-model failures while wrong_file dominates low-capability failures, and we derive practical tier-aware harness selection guidelines.

Related papers

• C$^3$PO: Evaluating Cross-Modal Composition and Counterfactual Performance in Omnimodal Models [11.992819604307998]
  We introduce C$3$PO, a benchmark of 3,404 samples spanning video, audio, image, and text.<n>We evaluate two abilities: information composition (fusing dispersed evidence) and counterfactual conflict.<n>Built through a fully automatic pipeline using 25 logically grounded templates, C$3$PO reveals that while humans achieve 88.64% accuracy, the best model reaches only 73.17%.
  arXiv  Detail & Related papers  (2026-08-05T20:04:05Z)
• Persona-Model Collapse in Emergent Misalignment [0.0]
  Fine-tuning large language models on narrow data with harmful content produces broadly misaligned behavior on unrelated prompts.<n>We test this hypothesis using two metrics: moral susceptibility (S) and moral robustness (R)<n>These metrics formalize the model's ability to differentiate characters (S) and its consistency when simulating a given one (R)
  arXiv  Detail & Related papers  (2026-05-13T00:48:57Z)
• The Model Says Walk: How Surface Heuristics Override Implicit Constraints in LLM Reasoning [9.898274894485107]
  Large language models systematically fail when a salient surface cue conflicts with an unstated feasibility constraint.<n>We study this through a diagnose-measure-bridge-treat framework.
  arXiv  Detail & Related papers  (2026-03-30T21:36:09Z)
• Silent Commitment Failure in Instruction-Tuned Language Models: Evidence of Governability Divergence Across Architectures [0.0]
  We introduce governability -- the degree to which a model's errors are detectable before output commitment and correctable once detected.<n>We show benchmark accuracy does not predict governability, correction capacity varies independently of detection, and identical governance scaffolds produce opposite effects across models.<n>We propose a Detection and Correction Matrix classifying model-task combinations into four regimes: Governable, Monitor Only, Steer Blind, and Ungovernable.
  arXiv  Detail & Related papers  (2026-03-22T21:50:28Z)
• Same Answer, Different Representations: Hidden instability in VLMs [65.36933543377346]
  We introduce a representation-aware and frequency-aware evaluation framework that measures internal embedding drift, spectral sensitivity, and structural smoothness.<n>We apply this framework to modern Vision Language Models (VLMs) across the SEEDBench, MMMU, and POPE datasets.
  arXiv  Detail & Related papers  (2026-02-06T12:24:26Z)
• The Devil in the Details: Emergent Misalignment, Format and Coherence in Open-Weights LLMs [0.0]
  We evaluate if current-generation open-weights models exhibit similar resistance to the Qwen-2.5 family.<n>We replicate the effect across nine modern open-weights models.
  arXiv  Detail & Related papers  (2025-11-25T09:25:33Z)
• Adapt in the Wild: Test-Time Entropy Minimization with Sharpness and Feature Regularization [85.50560211492898]
  Test-time adaptation (TTA) may fail to improve or even harm the model performance when test data have mixed distribution shifts.<n>This is often a key obstacle preventing existing TTA methods from being deployed in the real world.<n>We propose a sharpness-aware and reliable entropy minimization method, called SAR, for stabilizing TTA from two aspects.
  arXiv  Detail & Related papers  (2025-09-05T10:03:00Z)
• Goedel-Prover-V2: Scaling Formal Theorem Proving with Scaffolded Data Synthesis and Self-Correction [95.91743732150233]
  Goedel-Prover-V2, a series of open-source language models, set a new state-of-the-art in automated theorem proving.<n>We generate synthetic tasks of increasing difficulty to train the model to master increasingly complex theorems.<n>Goedel-Prover-V2-32B achieves 88.1% on MiniF2F at pass@32 in standard mode and 90.4% in self-correction mode.
  arXiv  Detail & Related papers  (2025-08-05T16:28:22Z)
• Multi-Level Collaboration in Model Merging [56.31088116526825]
  This paper explores the intrinsic connections between model merging and model ensembling.<n>We find that even when previous restrictions are not met, there is still a way for model merging to attain a near-identical and superior performance similar to that of ensembling.
  arXiv  Detail & Related papers  (2025-03-03T07:45:04Z)
• Are Sample-Efficient NLP Models More Robust? [90.54786862811183]
  We investigate the relationship between sample efficiency (amount of data needed to reach a given ID accuracy) and robustness (how models fare on OOD evaluation) We find that higher sample efficiency is only correlated with better average OOD robustness on some modeling interventions and tasks, but not others. These results suggest that general-purpose methods for improving sample efficiency are unlikely to yield universal OOD robustness improvements, since such improvements are highly dataset- and task-dependent.
  arXiv  Detail & Related papers  (2022-10-12T17:54:59Z)
• Sparse MoEs meet Efficient Ensembles [49.313497379189315]
  We study the interplay of two popular classes of such models: ensembles of neural networks and sparse mixture of experts (sparse MoEs) We present Efficient Ensemble of Experts (E$3$), a scalable and simple ensemble of sparse MoEs that takes the best of both classes of models, while using up to 45% fewer FLOPs than a deep ensemble.
  arXiv  Detail & Related papers  (2021-10-07T11:58:35Z)
• AvgOut: A Simple Output-Probability Measure to Eliminate Dull Responses [97.50616524350123]
  We build dialogue models that are dynamically aware of what utterances or tokens are dull without any feature-engineering. The first model, MinAvgOut, directly maximizes the diversity score through the output distributions of each batch. The second model, Label Fine-Tuning (LFT), prepends to the source sequence a label continuously scaled by the diversity score to control the diversity level. The third model, RL, adopts Reinforcement Learning and treats the diversity score as a reward signal.
  arXiv  Detail & Related papers  (2020-01-15T18:32:06Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.