{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,30]],"date-time":"2025-12-30T06:42:05Z","timestamp":1767076925374,"version":"3.48.0"},"reference-count":33,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,11,15]],"date-time":"2025-11-15T00:00:00Z","timestamp":1763164800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,11,15]],"date-time":"2025-11-15T00:00:00Z","timestamp":1763164800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,11,15]]},"DOI":"10.1109\/ipccc66453.2025.11304643","type":"proceedings-article","created":{"date-parts":[[2025,12,29]],"date-time":"2025-12-29T18:36:35Z","timestamp":1767033395000},"page":"1-8","source":"Crossref","is-referenced-by-count":0,"title":["Adaptive LLM Routing for Scientific Workflows: Predicting Query Complexity to Optimize Cost"],"prefix":"10.1109","author":[{"given":"Jonas","family":"Thierfeldt","sequence":"first","affiliation":[{"name":"Technische Universität Berlin,Berlin,Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Dominik","family":"Scheinert","sequence":"additional","affiliation":[{"name":"Technische Universität Berlin,Berlin,Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Thorsten","family":"Wittkopp","sequence":"additional","affiliation":[{"name":"Technische Universität Berlin,Berlin,Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Odej","family":"Kao","sequence":"additional","affiliation":[{"name":"Technische Universität Berlin,Berlin,Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.emnlp-main.895"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1109\/CCGrid57682.2023.00025"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1093\/bib\/bbac409"},{"key":"ref4","article-title":"Galactica: A large language model for science","author":"Taylor","year":"2022","journal-title":"CoRR"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.207"},{"key":"ref6","article-title":"Mapping the challenges of HCI: an application and evaluation of chatgpt and GPT-4 for cost-efficient question answering","author":"Oppenlaender","year":"2023","journal-title":"CoRR"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3624062.3624172"},{"key":"ref8","doi-asserted-by":"crossref","DOI":"10.1145\/3458817.3476209","article-title":"Efficient large-scale language model training on GPU clusters using megatron-lm","volume-title":"SC","author":"Narayanan","year":"2021"},{"key":"ref9","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"NeurIPS"},{"key":"ref10","article-title":"Flashattention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2024","journal-title":"ICLR. OpenReview.net"},{"key":"ref11","article-title":"Quality time: Carbon-aware quality adaptation for energy-intensive services","author":"Wiesner","year":"2024","journal-title":"CoRR"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.204"},{"key":"ref13","article-title":"Towards optimal caching and model selection for large model inference","author":"Zhu","year":"2023","journal-title":"NeurIPS"},{"key":"ref14","article-title":"Cascade speculative drafting for even faster LLM inference","author":"Chen","year":"2024","journal-title":"NeurIPS"},{"key":"ref15","article-title":"Bartscore: Evaluating generated text as text generation","author":"Yuan","year":"2021","journal-title":"NeurIPS"},{"key":"ref16","article-title":"Complexitynet: Increasing LLM inference efficiency by learning task complexity","author":"Bae","year":"2023","journal-title":"CoRR"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.findings-acl.108"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.21236\/ada006655"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.nlppower-1.6"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.297"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.naacl-long.389"},{"key":"ref22","article-title":"Fly-swat or cannon? cost-effective language model choice via meta-modeling","author":"Sakota","year":"2024","journal-title":"WSDM. ACM"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/ICWS62655.2024.00098"},{"key":"ref24","article-title":"Large language model routing with benchmark datasets","author":"Shnitzer","year":"2023","journal-title":"CoRR"},{"key":"ref25","article-title":"Hybrid LLM: cost-efficient and quality-aware query routing","author":"Ding","year":"2024","journal-title":"ICLR. OpenReview.net"},{"key":"ref26","article-title":"Routellm: Learning to route 11 ms with preference data","author":"Ong","year":"2024","journal-title":"CoRR"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-industry.34"},{"key":"ref28","article-title":"Lightrouter: Towards efficient 11 m collaboration with minimal overhead","author":"Zhang","year":"2025","journal-title":"CoRR"},{"key":"ref29","article-title":"CITER: collaborative inference for efficient large language model decoding with token-level routing","author":"Zheng","year":"2025","journal-title":"CoRR"},{"key":"ref30","article-title":"Universal model routing for efficient LLM inference","author":"Jitkrittum","year":"2025","journal-title":"CoRR"},{"key":"ref31","article-title":"Eagle: Efficient training-free router for multi-llm inference","author":"Zhao","year":"2024","journal-title":"CoRR"},{"key":"ref32","article-title":"Routereval: A comprehensive benchmark for routing 11 ms to explore model-level scaling up in 11 ms","author":"Huang","year":"2025","journal-title":"CoRR"},{"key":"ref33","article-title":"Rethinking predictive modeling for 11 m routing: When simple knn beats complex learned routers","author":"Li","year":"2025","journal-title":"CoRR"}],"event":{"name":"2025 IEEE International Performance, Computing, and Communications Conference (IPCCC)","start":{"date-parts":[[2025,11,15]]},"location":"Austin, TX, USA","end":{"date-parts":[[2025,11,23]]}},"container-title":["2025 IEEE International Performance, Computing, and Communications Conference (IPCCC)"],"original-title":[],"link":[{"URL":"https:\/\/2.zoppoz.workers.dev:443\/http\/xplorestaging.ieee.org\/ielx8\/11304598\/11304625\/11304643.pdf?arnumber=11304643","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,30]],"date-time":"2025-12-30T06:39:39Z","timestamp":1767076779000},"score":1,"resource":{"primary":{"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/ieeexplore.ieee.org\/document\/11304643\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,11,15]]},"references-count":33,"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/doi.org\/10.1109\/ipccc66453.2025.11304643","relation":{},"subject":[],"published":{"date-parts":[[2025,11,15]]}}}