mlsanityData to 14 Oct 2026

Jumping the Line: Exploiting Length Predictions in LLM Scheduling

arXiv 2610.03430 ↗First posted 2 Oct 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Efficient request scheduling is increasingly important for reducing completion time in large language model (LLM) serving. Size-based policies such as Shortest Job First prioritize shorter requests, but output lengths are unknown before generation, so practical schedulers rely on predicted lengths. We introduce JIL, an attack on prediction-based LLM schedulers that manipulates the scheduling signal to obtain higher priority and reduce completion time. Using TRAIL as a case study, JIL optimizes an adversarial suffix that causes a lightweight output-length probe to underestimate a request's length. We evaluate JIL on two datasets and four LLMs across varied request profiles and deployment configurations. JIL reduces predicted output lengths by up to 83.4 percent, and adversarial requests complete up to 1.53 times faster on average in end-to-end serving experiments. The reduction in predicted length is substantially larger than the change in actual output length, revealing a mismatch between the scheduler's estimate and the request's realized size. Response utility varies across models and tasks, exposing a trade-off between scheduling advantage and response quality. We also evaluate scheduler-side defenses and find that grouping length predictions into coarse intervals reduces JIL's scheduling advantage and mitigates delays to benign requests.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 31 of its 36 references that are in the map. First posted on arXiv 2 Oct 2026.

Builds on

31 of 36 references.

References · 31

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  4. 2021Probing Classifiers: Promises, Shortcomings, and Advancescited by 876
  5. 2023AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Modelscited by 785
  6. 2024DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Servingcited by 617
  7. 2023Splitwise: Efficient generative LLM inference using phase splittingcited by 558
  8. 2024Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Servecited by 526
  9. 2023SmoothLLM: Defending Large Language Models Against Jailbreaking Attackscited by 469
  10. 2023Using an LLM to Help With Code Understandingcited by 228
  11. 2024Llumnix: Dynamic Scheduling for Large Language Model Servingcited by 174
  12. 2024Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attackscited by 156
19 more
  1. 2020Sponge Examples: Energy-Latency Attacks on Neural Networkscited by 111
  2. 2024Efficient LLM Scheduling by Learning to Rankcited by 85
  3. 2024Defending LLMs against Jailbreaking Attacks via Backtranslationcited by 70
  4. 2024Efficient Interactive LLM Serving with Proxy Model-based Sequence Length Predictioncited by 66
  5. 2024Efficiently Scaling LLM Reasoning with Certaindexcited by 58
  6. 2025JBShield: Defending Large Language Models from Jailbreak Attacks through Activated Concept Analysis and Manipulationcited by 54
  7. 2019Scheduling with Predictions and the Price of Mispredictioncited by 47
  8. 2024Don't Stop Me Now: Embedding Based Scheduling for LLMscited by 35
  9. 2024Denial-of-Service Poisoning Attacks against Large Language Modelscited by 35
  10. 2024Crabs: Consuming Resource via Auto-generation for LLM-DoS Attack under Black-box Settingscited by 35
  11. 2025ThinkTrap: Denial-of-Service Attacks against Black-box LLM Services via Infinite Thinkingcited by 23
  12. 2022Scalable Verification of GNN-based Job Schedulerscited by 19
  13. 2025Universal Jailbreak Suffixes Are Strong Attention Hijackerscited by 10
  14. 2025ELIS: Efficient LLM Iterative Scheduling System with Response Length Predictorcited by 9
  15. 2025LoopLLM: Transferable Energy-Latency Attacks in LLMs via Repetitive Generationcited by 8
  16. 2026Scheduling LLM Inference with Uncertainty-Aware Output Length Predictionscited by 8
  17. 2026Predictive Scheduling for Efficient Inference-Time Reasoning in Large Language Modelscited by 3
  18. 2025Intra-request branch orchestration for efficient LLM reasoningcited by 1
  19. 2026OUTLETS: Output-Length Prediction from Speculative Decoding Backbonescited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024ScaleLLM: A Resource-Frugal LLM Serving Framework by Optimizing End-to-End Efficiency
  2. 2025Optimal Scheduling Algorithms for LLM Inference: Theory and Practice
  3. 2025Ranking Before Serving: Low-Latency LLM Serving via Pairwise Learning-to-Rank
  4. 2026Beyond Prediction: Tail-Aware Scheduling for LLM Inference
  5. 2026Cascade: Exploiting SLO-Aware latency budget for fair and high goodput LLM inference serving
  6. 2023Response Length Perception and Sequence Scheduling: An LLM-Empowered LLM Inference Pipeline
  7. 2025LLM Serving Optimization with Variable Prefill and Decode Lengths
  8. 2025CascadeInfer: Length-Aware Scheduling of LLM Serving with Low Latency and Load Balancing