cs.AINov 16, 2026

The Logic Chain Breaking Point: A Controlled-Variable Benchmark for Reasoning Degradation in Open-Weight Language Models

Authors: Akif Islam, Mohd. Ruhul Ameen, Md. Ekramul Hamid

Organizations: Department of Computer Science and Engineering, University of Rajshahi, Rajshahi, 6205, Bangladesh · Department of Computer Science and Engineering, Varendra University, Rajshahi, Bangladesh

Abstract

We introduce The Logic Chain Breaking Point, a controlled-variable benchmark for deductive reasoning in open-weight language models. It spans three phases: (i) pure modus-ponens chain scaling, (ii) distractor resilience with a fixed core chain, and (iii) contradiction detection under primacy versus recency placement. Independent variables follow a powers-of-two sequence v∈{1,2,4,8,16,32,64,128}v \in \{1,2,4,8,16,32,64,128\}, with ground truth known by construction. We evaluate sixteen open-weight models across instruction-tuned and reasoning-tuned panels, including a small-model suite and a larger local-model extension. All models are evaluated with greedy decoding under model-appropriate context and output budgets, while context- and length-limited trials are quarantined from the valid accuracy denominator. Our results reveal a sharply non-monotonic difficulty surface: pure deductive chains up to L=128L=128 remain near saturation for most models, and distractor noise causes only model-specific degradation at high DD, while the inconsistent premise condition has the lowest accuracy. Instruction-tuned models, including larger 24B--27B baselines, score essentially zero on inconsistent premise sets, while reasoning-tuned models recover only partially and remain below reliable accuracy. Within this benchmark, inconsistent premise sets are associated with the strongest degradation under the benchmark's abstention-based scoring rule.

Figures & tables

Explore similar work

Apr 8, 2026

Unified Deployment-Aware Evaluation of Open Reasoning Language Models

Open reasoning language models are often compared under mixed sample sizes, partially standardized prompts, and accuracy-centered summaries, which makes practical model selection difficult to interpret. We present a unified evaluation of seven open reasoning language model configurations across four benchmarks: ARC-Challenge, GSM8K, MATH levels 1 to 3, and TruthfulQA MC1. We test zero-shot, chain-of-thought (CoT), and few-shot CoT prompting on the same 238-example subset for every model--dataset--strategy condition, yielding a complete 7 x 4 x 3 design with 84 conditions and 19,992 evaluated examples. Beyond accuracy, we report Wilson confidence intervals, latency, peak video random access memory (VRAM), weighted aggregate performance, Pareto-efficient operating points, prompt-sensitivity metrics, and compatibility diagnostics. Gemma-4-26B-A4B with zero-shot prompting achieves the highest weighted score at 0.794. Gemma-4-E4B remains close to the top across prompting settings while using substantially lower latency and memory, making it a strong practical operating point. Bootstrap and paired-permutation analyses show that the leading configurations are close enough that deployment tradeoffs remain important. We also find that prompting strategy changes model rankings rather than shifting all models uniformly. Benchmark-specific complementarity creates routing headroom, with an oracle task-aware selector reaching a weighted score of 0.825. Compatibility diagnostics show that some apparent failures, especially Phi-4-Reasoning on GSM8K, reflect robustness and interface-adherence problems under the shared evaluation pipeline. These results support a central claim: open-model evaluation should be framed as a deployment-aware, multi-objective operating-point problem rather than as a single-score leaderboard exercise.
Mar 23, 2026

Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models?

Chain-of-thought (CoT) reasoning has been proposed as a transparency mechanism for large language models in safety-critical deployments, yet its effectiveness depends on faithfulness (whether models accurately verbalize the factors that actually influence their outputs), a property that prior evaluations have examined in only two proprietary models, finding acknowledgment rates as low as 25% for Claude 3.7 Sonnet and 39% for DeepSeek-R1. To extend this evaluation across the open-weight ecosystem, this study tests 12 open-weight reasoning models spanning 9 architectural families (7B-685B parameters) on 498 multiple-choice questions from MMLU and GPQA Diamond, injecting six categories of reasoning hints (sycophancy, consistency, visual pattern, metadata, grader hacking, and unethical information) and measuring the rate at which models acknowledge hint influence in their CoT when hints successfully alter answers. Across 41,832 inference runs, overall faithfulness rates range from 39.7% (Seed-1.6-Flash) to 89.9% (DeepSeek-V3.2-Speciale) across model families, with consistency hints (35.5%) and sycophancy hints (53.9%) exhibiting the lowest acknowledgment rates. Training methodology and model family predict faithfulness more strongly than parameter count, and keyword-based analysis reveals a striking gap between thinking-token acknowledgment (approximately 87.5%) and answer-text acknowledgment (approximately 28.6%), suggesting that models internally recognize hint influence but systematically suppress this acknowledgment in their outputs. These findings carry direct implications for the viability of CoT monitoring as a safety mechanism and suggest that faithfulness is not a fixed property of reasoning models but varies systematically with architecture, training method, and the nature of the influencing cue.
Feb 16, 2026

Broken Chains: The Cost of Incomplete Reasoning in LLMs

Reasoning-specialized models like OpenAI's 5.1 and DeepSeek-V3.2 allocate substantial inference compute to extended chain-of-thought (CoT) traces, yet reasoning tokens incur significant costs. How do different reasoning modalities of code, natural language, hybrid, or none do perform under token constraints? We introduce a framework that constrains models to reason exclusively through code, comments, both, or neither, then systematically ablates token budgets to 10%, 30%, 50%, and 70% of optimal. We evaluate four frontier models (GPT-5.1, Gemini 3 Flash, DeepSeek-V3.2, Grok 4.1) across mathematical benchmarks (AIME, GSM8K, HMMT). Our findings reveal: (1) truncated reasoning can hurt as DeepSeek-V3.2 achieves 53% with no reasoning but only 17% with truncated CoT at 50% budget; (2) code degrades gracefully as Gemini's comments collapse to 0% while code maintains 43-47%; (3) hybrid reasoning underperforms single modalities; (4) robustness is model-dependent as Grok maintains 80-90% at 30% budget where OpenAI and DeepSeek collapse to 7-27%. These results suggest incomplete reasoning chains actively mislead models, with implications for deploying reasoning-specialized systems under resource constraints.