{"paper":{"title":"Unified Deployment-Aware Evaluation of Open Reasoning Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Accuracy-efficiency tradeoffs in reasoning LLMs depend jointly on architecture, prompting protocol, and task rather than sparse activation alone.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ge Wang, Md Motaleb Hossen Manik","submitted_at":"2026-04-08T12:50:52Z","abstract_excerpt":"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 e"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"These results show that sparse activation alone does not guarantee the best practical operating point: observed accuracy-efficiency tradeoffs depend jointly on architecture, prompting protocol, and task composition.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the four chosen benchmarks and three prompting strategies are representative enough of real-world reasoning workloads to support general claims about accuracy-efficiency tradeoffs.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Gemma-4-E4B with few-shot chain-of-thought reaches the highest weighted accuracy of 0.675 at 14.9 GB VRAM, while the larger Gemma-4-26B-A4B MoE model scores 0.663 but uses 48.1 GB.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Accuracy-efficiency tradeoffs in reasoning LLMs depend jointly on architecture, prompting protocol, and task rather than sparse activation alone.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"455f26886c7e9c24405dc6418b419b58d2eafe09b327f5aac3e5575be5729663"},"source":{"id":"2604.07035","kind":"arxiv","version":2},"verdict":{"id":"93e4922e-cb2b-4db0-bdc4-28d92264b317","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T17:44:59.556352Z","strongest_claim":"These results show that sparse activation alone does not guarantee the best practical operating point: observed accuracy-efficiency tradeoffs depend jointly on architecture, prompting protocol, and task composition.","one_line_summary":"Gemma-4-E4B with few-shot chain-of-thought reaches the highest weighted accuracy of 0.675 at 14.9 GB VRAM, while the larger Gemma-4-26B-A4B MoE model scores 0.663 but uses 48.1 GB.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the four chosen benchmarks and three prompting strategies are representative enough of real-world reasoning workloads to support general claims about accuracy-efficiency tradeoffs.","pith_extraction_headline":"Accuracy-efficiency tradeoffs in reasoning LLMs depend jointly on architecture, prompting protocol, and task rather than sparse activation alone."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.07035/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}