Green Decoding: ELQ Co-Optimization and Carbon-Aware Scheduling for Efficient LLM Inference

Green Decoding: ELQ Co-Optimization and Carbon-Aware Scheduling for Efficient LLM Inference | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Green Decoding: ELQ Co-Optimization and Carbon-Aware Scheduling for Efficient LLM Inference Gaith Rjoub, Jamal Bentahar, Shahed Almobydeen, Ayoub Alsarhan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9252320/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract The large-scale deployment of Large Language Models (LLMs) is constrained by significant energy consumption and operational costs, with inference accounting for up to 90% of the total energy footprint. Existing optimization methods typically address latency or memory independently, frequently overlooking energy efficiency, carbon impact, and their intricate relationship with user-defined Service Level Agreements (SLAs) for quality and response time. This work presents Green Decoding, a novel co-optimization framework for LLM inference. Green Decoding formulates inference as a multi-objective optimization problem, minimizing a weighted function of Energy, Latency, and Quality (ELQ). The framework utilizes a policy engine that, on a per-request basis, jointly tunes a broad set of system parameters, including speculative decoding configurations, dynamic Key-Value (KV) cache policies, adaptive quantization tiers, and early-exit criteria. The framework introduces two key contributions: (1) a carbon-aware scheduler that leverages real-time grid carbon intensity data to strategically time-shift deferrable, non-interactive workloads to periods of cleaner energy, thereby directly reducing CO 2 emissions without violating SLAs, and (2) 1 a safety-aware gating mechanism that employs runtime uncertainty and toxicity signals to limit aggressive, potentially quality-degrading optimizations, thereby ensuring model reliability. Across diverse workloads, Green Decoding demonstrates superior performance, establishing a more efficient ELQ Pareto frontier. The framework achieves up to 35% energy reduction and 50% lower carbon emissions (gCO 2 e) compared to highly optimized static baselines, while strictly adhering to p95 latency and quality-proxy SLAs. Large Language Models (LLMs) Energy-Latency-Quality (ELQ) Optimization Carbon-Aware Scheduling Green AI / Sustainable AI Efficient LLM Inference Adaptive Inference Systems Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 11 May, 2026 Reviews received at journal 09 May, 2026 Reviewers agreed at journal 04 May, 2026 Reviewers agreed at journal 28 Apr, 2026 Reviewers invited by journal 15 Apr, 2026 Editor assigned by journal 30 Mar, 2026 Submission checks completed at journal 30 Mar, 2026 First submitted to journal 28 Mar, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9252320","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":624680185,"identity":"de5810e3-c478-40cb-89c6-1a4965473a77","order_by":0,"name":"Gaith Rjoub","email":"","orcid":"","institution":"Aqaba University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Gaith","middleName":"","lastName":"Rjoub","suffix":""},{"id":624680187,"identity":"f377c691-d13e-45a9-be93-e3c47617c0df","order_by":1,"name":"Jamal Bentahar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCElEQVRIiWNgGAWjYBACxgYg8YDhAIjNBsQ2BsRpSQBpYQNrSSOsBQyQtBwmrIW5/XTihwSGO3Lm8s3PHvzcc97Y4ADzww8MNTa4HdaTu1kigeGZsWUbm7lhz7PbZgYH2IwlGI6l4fFL7gaglsOJG44xmEnwHLhtY3CAwQzozMO4tfS/3fwDqKV+wzH2b5J/DpwDamH/xsDw7z9uLTNyt4FsSTA4xmMmzXPgANBhPGYMjG0H8Gh5u80iweCZ4YZjOWXSMgeSjSUP8xRLJPYl49Ri2J+7+caHijvyBoePb5N8c8DOsO94+8YPH77Z4dbSACJRIoOZARRTuIE8HrlRMApGwSgYBRAAAANGWGbhh6rBAAAAAElFTkSuQmCC","orcid":"","institution":"Khalifa University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Jamal","middleName":"","lastName":"Bentahar","suffix":""},{"id":624680189,"identity":"259d4590-dbdc-49e7-91d1-7a24ec80effc","order_by":2,"name":"Shahed Almobydeen","email":"","orcid":"","institution":"Aqaba University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Shahed","middleName":"","lastName":"Almobydeen","suffix":""},{"id":624680191,"identity":"791ef832-8780-4e47-8feb-398dba1c0ee0","order_by":3,"name":"Ayoub Alsarhan","email":"","orcid":"","institution":"Hashemite University","correspondingAuthor":false,"prefix":"","firstName":"Ayoub","middleName":"","lastName":"Alsarhan","suffix":""}],"badges":[],"createdAt":"2026-03-28 11:38:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9252320/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9252320/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107420809,"identity":"bef6c17d-5a68-4426-9e57-91a3cd5aee5a","added_by":"auto","created_at":"2026-04-21 10:26:40","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4206045,"visible":true,"origin":"","legend":"","description":"","filename":"GreenDecodingEvolutionaryIntelligence20261.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9252320/v1_covered_e504632e-fdc4-4f01-a429-efaae6029bcf.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Green Decoding: ELQ Co-Optimization and Carbon-Aware Scheduling for Efficient LLM Inference","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"evolutionary-intelligence","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"evin","sideBox":"Learn more about [Evolutionary Intelligence](http://link.springer.com/journal/12065)","snPcode":"12065","submissionUrl":"https://submission.nature.com/new-submission/12065/3","title":"Evolutionary Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Large Language Models (LLMs), Energy-Latency-Quality (ELQ) Optimization, Carbon-Aware Scheduling, Green AI / Sustainable AI, Efficient LLM Inference, Adaptive Inference Systems ","lastPublishedDoi":"10.21203/rs.3.rs-9252320/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9252320/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe large-scale deployment of Large Language Models (LLMs) is constrained by significant energy consumption and operational costs, with inference accounting for up to 90% of the total energy footprint. Existing optimization methods typically address latency or memory independently, frequently overlooking energy efficiency, carbon impact, and their intricate relationship with user-defined Service Level Agreements (SLAs) for quality and response time. This work presents Green Decoding, a novel co-optimization framework for LLM inference. Green Decoding formulates inference as a multi-objective optimization problem, minimizing a weighted function of Energy, Latency, and Quality (ELQ). The framework utilizes a policy engine that, on a per-request basis, jointly tunes a broad set of system parameters, including speculative decoding configurations, dynamic Key-Value (KV) cache policies, adaptive quantization tiers, and early-exit criteria. The framework introduces two key contributions: (1) a carbon-aware scheduler that leverages real-time grid carbon intensity data to strategically time-shift deferrable, non-interactive workloads to periods of cleaner energy, thereby directly reducing CO\u003csub\u003e2\u003c/sub\u003e emissions without violating SLAs, and (2) 1 a safety-aware gating mechanism that employs runtime uncertainty and toxicity signals to limit aggressive, potentially quality-degrading optimizations, thereby ensuring model reliability. Across diverse workloads, Green Decoding demonstrates superior performance, establishing a more efficient ELQ Pareto frontier. The framework achieves up to 35% energy reduction and 50% lower carbon emissions (gCO\u003csub\u003e2\u003c/sub\u003ee) compared to highly optimized static baselines, while strictly adhering to p95 latency and quality-proxy SLAs.\u003c/p\u003e","manuscriptTitle":"Green Decoding: ELQ Co-Optimization and Carbon-Aware Scheduling for Efficient LLM Inference","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-21 10:23:03","doi":"10.21203/rs.3.rs-9252320/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-11T06:26:00+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-09T11:43:05+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"183477456430025630356806790062905240595","date":"2026-05-04T04:25:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"153654814532158360222509154704553013558","date":"2026-04-29T03:42:47+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-15T04:28:19+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-30T05:27:50+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-30T05:27:22+00:00","index":"","fulltext":""},{"type":"submitted","content":"Evolutionary Intelligence","date":"2026-03-28T11:30:39+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"evolutionary-intelligence","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"evin","sideBox":"Learn more about [Evolutionary Intelligence](http://link.springer.com/journal/12065)","snPcode":"12065","submissionUrl":"https://submission.nature.com/new-submission/12065/3","title":"Evolutionary Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"d3bb3766-43a4-4b89-8c74-f2e0ab5a24a9","owner":[],"postedDate":"April 21st, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-11T06:26:00+00:00","index":36,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-09T11:43:05+00:00","index":35,"fulltext":""},{"type":"reviewerAgreed","content":"183477456430025630356806790062905240595","date":"2026-05-04T04:25:11+00:00","index":33,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-21T10:23:04+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-21 10:23:03","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9252320","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9252320","identity":"rs-9252320","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}