{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T16:19:14Z","timestamp":1783700354589,"version":"3.55.0"},"reference-count":67,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,12,8]],"date-time":"2025-12-08T00:00:00Z","timestamp":1765152000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,12,8]],"date-time":"2025-12-08T00:00:00Z","timestamp":1765152000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,12,8]]},"DOI":"10.1109\/bigdata66926.2025.11400957","type":"proceedings-article","created":{"date-parts":[[2026,3,6]],"date-time":"2026-03-06T20:57:57Z","timestamp":1772830677000},"page":"2234-2239","source":"Crossref","is-referenced-by-count":3,"title":["A Cost-Benefit Analysis of On-Premise Large Language Model Deployment: Breaking Even with Commercial LLM Services"],"prefix":"10.1109","author":[{"given":"Guanzhong","family":"Pan","sequence":"first","affiliation":[{"name":"Carnegie Mellon University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Vishal","family":"Chodnekar","sequence":"additional","affiliation":[{"name":"Unaffiliated"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Abinas","family":"Roy","sequence":"additional","affiliation":[{"name":"Unaffiliated"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Haibo","family":"Wang","sequence":"additional","affiliation":[{"name":"Carnegie Mellon University"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/icssas64001.2024.10760403"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.3390\/info16080688"},{"key":"ref3","article-title":"A Survey of Large Language Models","volume-title":"arXiv","author":"Zhao","year":"2023"},{"key":"ref4","doi-asserted-by":"crossref","DOI":"10.2139\/ssrn.5296479","article-title":"Cloud or On-Premise? A Strategic View of Large Language Model Deployment","volume-title":"SSRN","author":"Zhang","year":"2025"},{"key":"ref5","article-title":"Position: On-Premises LLM Deployment Demands a Middle Path: Preserving Privacy Without Sacrificing Model Confidentiality","volume-title":"arXiv, 2024","author":"Huang","year":"2025"},{"key":"ref6","author":"Luqman","year":"2025","journal-title":"Agentic AI in Finance: A Comprehensive Overview"},{"key":"ref7","article-title":"FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance","volume-title":"arXiv, 2023","author":"Chen","year":"2025"},{"key":"ref8","doi-asserted-by":"crossref","DOI":"10.1007\/s44443-025-00177-1","article-title":"A Survey on Privacy Risks and Protection in Large Language Models","volume-title":"arXiv, 2025","author":"Chen","year":"2025"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01476-7"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/BigData62323.2024.10825658"},{"key":"ref11","article-title":"Llama 2: Open Foundation and Fine-Tuned Chat Models","author":"Touvron","year":"2023","journal-title":"arXiv preprint"},{"key":"ref12","article-title":"Mistral 7B","author":"Jiang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref13","article-title":"Qwen Technical Report","author":"Bai","year":"2023","journal-title":"arXiv preprint"},{"key":"ref14","article-title":"NVIDIA H100 Tensor Core GPU Architecture Overview","volume-title":"NVIDIA","year":"2025"},{"key":"ref15","article-title":"AMD Instinct MI300X Accelerators","volume-title":"AMD","year":"2025"},{"key":"ref16","article-title":"Efficient Memory Management for Large Language Model Serving with PagedAttention","volume-title":"arXiv, 2023","author":"Kwon","year":"2024"},{"key":"ref17","article-title":"NVIDIA TensorRT-LLM","volume-title":"NVIDIA Documentation, 2024","year":"2025"},{"key":"ref18","article-title":"DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale","volume-title":"arXiv, 2022","author":"Aminabadi","year":"2025"},{"key":"ref19","article-title":"A Survey on Inference Engines for Large Language Models: Perspectives on Optimization and Efficiency","volume-title":"arXiv.org, 2025","author":"Park","year":"2025"},{"key":"ref20","article-title":"A Survey of Large Language Models","author":"Zhao","year":"2023","journal-title":"arXiv (Cornell University)"},{"key":"ref21","article-title":"Membership Inference Attacks on Large-Scale Models: A Survey","volume-title":"arXiv.org, 2025","author":"Wu","year":"2025"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1145\/3662006.3662067"},{"key":"ref23","doi-asserted-by":"crossref","DOI":"10.18653\/v1\/2025.naacl-long.328","article-title":"Decoding Speculative Decoding","volume-title":"arXiv.org, 2024","author":"Yan","year":"2025"},{"key":"ref24","article-title":"DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale","volume-title":"arXiv.org, 2022","author":"Aminabadi","year":"2024"},{"key":"ref25","article-title":"FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU","volume-title":"arXiv.org, 2023","author":"Sheng","year":"2025"},{"key":"ref26","article-title":"Fast Inference from Transformers via Speculative Decoding","volume-title":"arXiv.org, 2022","author":"Leviathan","year":"2025"},{"key":"ref27","article-title":"ServerlessLLM: Low-Latency Serverless Inference for Large Language Models","volume-title":"arXiv.org","author":"Fu","year":"2024"},{"key":"ref28","article-title":"Provisioned Throughput Units - Azure OpenAI","volume-title":"Microsoft","year":"2025"},{"key":"ref29","article-title":"S-LoRA: Serving Thousands of Concurrent LoRA Adapters","volume-title":"MLSys","author":"Sheng","year":"2024"},{"key":"ref30","article-title":"Accelerating Serverless LLM Inference with Materialization (Medusa)","author":"Zhao","year":"2025","journal-title":"SoCC"},{"key":"ref31","article-title":"Inference Economics of Language Models","volume-title":"arXiv.org, 2025","author":"Erdil","year":"2025"},{"key":"ref32","article-title":"An Inquiry into Datacenter TCO for LLM Inference with FP8","volume-title":"arXiv.org, 2025","author":"Kim","year":"2025"},{"key":"ref33","doi-asserted-by":"crossref","DOI":"10.18653\/v1\/2025.acl-long.1563","article-title":"Energy Considerations of Large Language Model Inference and Efficiency Optimizations","volume-title":"arXiv.org, 2025","author":"Fernandez","year":"2025"},{"key":"ref34","article-title":"LLaMA: Open and Efficient Foundation Language Models","author":"Touvron","year":"2023","journal-title":"arXiv"},{"key":"ref35","article-title":"Qwen Technical Report","volume-title":"arXiv.org","author":"Bai","year":"2023"},{"key":"ref36","article-title":"Kimi K2: Open Agentic Intelligence","author":"Team","year":"2025","journal-title":"arXiv.org, 2025"},{"key":"ref37","article-title":"moonshotai\/Kimi-K2-Instruct \u2022 Hugging Face","volume-title":"Huggingface.co","year":"2025"},{"key":"ref38","article-title":"GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models","year":"2025","journal-title":"arXiv.org, 2025"},{"key":"ref39","article-title":"zai-org\/GLM-4.5 \u2022 Hugging Face","volume-title":"Huggingface.co","year":"2025"},{"key":"ref40","article-title":"Qwen3 Technical Report","volume-title":"arXiv.org","author":"Yang","year":"2025"},{"key":"ref41","article-title":"Qwen\/Qwen3\u2013235B-A22B \u2022 Hugging Face","year":"2025","journal-title":"Huggingface.co"},{"key":"ref42","article-title":"gpt-oss-120b & gpt-oss-20b Model Card","volume-title":"arXiv.org, 2025","year":"2025"},{"key":"ref43","article-title":"openai\/gpt-oss-120b \u2022 Hugging Face","volume-title":"Huggingface.co","year":"2025"},{"key":"ref44","article-title":"Blocked","volume-title":"Huggingface.co, 2025","year":"2025"},{"key":"ref45","article-title":"The Llama 3 Herd of Models","volume-title":"arXiv.org","author":"Dubey","year":"2024"},{"key":"ref46","article-title":"meta-llama\/Llama-3.3\u201370B-Instruct \u2022 Hugging Face","year":"2024","journal-title":"Huggingface.co"},{"key":"ref47","article-title":"EXAONE 4.0: Unified Large Language Models Integrating Non-reasoning and Reasoning Modes","author":"AI","year":"2025","journal-title":"arXiv.org, 2025"},{"key":"ref48","article-title":"LGAI-EXAONE\/EXAONE-4.0\u201332B \u2022 Hugging Face","year":"2025","journal-title":"Huggingface.co"},{"key":"ref49","article-title":"Qwen\/Qwen3\u201330B-A3B \u2022 Hugging Face","year":"2025","journal-title":"Huggingface.co"},{"key":"ref50","article-title":"Magistral","volume-title":"arXiv.org, 2025","year":"2025"},{"key":"ref51","article-title":"mistralai\/Magistral-Small-2506 \u2022 Hugging Face","volume-title":"Huggingface.co, 2025","year":"2025"},{"key":"ref52","volume-title":"API Pricing","year":"2024"},{"key":"ref53","volume-title":"Claude Pricing","year":"2024"},{"key":"ref54","article-title":"Models and Pricing - xAI Docs","volume-title":"Docs.x.ai","year":"2024"},{"key":"ref55","article-title":"Gemini Developer API Pricing","volume-title":"Google AI for Developers","year":"2025"},{"key":"ref56","article-title":"An Inquiry into Datacenter TCO for LLM Inference with FP8","volume-title":"arXiv.org","author":"Kim","year":"2025"},{"key":"ref57","article-title":"NVIDIA GeForce RTX 5090 Graphics Cards","volume-title":"NVIDIA","year":"2025"},{"key":"ref58","article-title":"NVIDIA A100 Tensor Core GPU Architecture: Unprecedented Acceleration at Every Scale","volume-title":"NVIDIA White Paper","year":"2020"},{"key":"ref59","volume-title":"LiveBench: A Challenging, Contamination-Free LLM Benchmark","year":"2024"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1109\/BigData62323.2024.10825845"},{"key":"ref61","article-title":"GPQA: A Graduate-Level GoogleProof Q&A Benchmark","volume-title":"arXiv","author":"Rein","year":"2023"},{"key":"ref62","article-title":"Measuring Mathematical Problem Solving With the MATH Dataset","volume-title":"arXiv:2103.03874 [cs]","author":"Hendrycks","year":"2021"},{"key":"ref63","article-title":"LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code","volume-title":"arXiv","author":"Jain","year":"2024"},{"key":"ref64","article-title":"Model & API Providers Analysis - Artificial Analysis","volume-title":"artificialanalysis.ai","author":"Analysis","year":"2025"},{"key":"ref65","article-title":"MMLU-Pro: A More Robust and Challenging MultiTask Language Understanding Benchmark","volume-title":"arXiv","author":"Wang","year":"2024"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1109\/BigData62323.2024.10825845"},{"key":"ref67","article-title":"AI Model & API Providers Analysis","volume-title":"Artificial Analysis","year":"2025"}],"event":{"name":"2025 IEEE International Conference on Big Data (BigData)","location":"Macau, China","start":{"date-parts":[[2025,12,8]]},"end":{"date-parts":[[2025,12,11]]}},"container-title":["2025 IEEE International Conference on Big Data (BigData)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11400704\/11400712\/11400957.pdf?arnumber=11400957","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,7]],"date-time":"2026-03-07T06:52:56Z","timestamp":1772866376000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11400957\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,8]]},"references-count":67,"URL":"https:\/\/doi.org\/10.1109\/bigdata66926.2025.11400957","relation":{},"subject":[],"published":{"date-parts":[[2025,12,8]]}}}