Paper Detail
Shuhuai Li, Shengdong Ye, Dongdong Ge, Yinyu Ye, Jianghao Lin
Agentic services alternate between model inference on clustered GPUs and external tools on distributed CPUs. In this scenario, existing systems (1) overlook tool-return readiness in service evaluation, wasting limited resources, and (2) use heuristic strategies to decide admission and placement on GPUs, resulting in suboptimal scheduling decisions. To bridge this gap, we present TAO, a tool-aware joint scheduling framework for LLM-based agentic serving. Its value interface combines near-term tool readiness, node-specific cache credit with time-aware decay, and token footprint; a dynamic program jointly selects and places waiting sessions under capacity constraints. Across the 12 reported model-workload-concurrency settings, TAO reduces latency by up to 26.8%, reduces mean time to first token (TTFT) by up to 36.3%, and increases decode throughput by up to 29.5% relative to the best evaluated state-of-the-art baseline for each metric.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{li2026tao,
title = {TAO: Tool-Aware Joint Scheduling for LLM-Based Agentic Serving},
author = {Shuhuai Li and Shengdong Ye and Dongdong Ge and Yinyu Ye and Jianghao Lin},
year = {2026},
abstract = {Agentic services alternate between model inference on clustered GPUs and external tools on distributed CPUs. In this scenario, existing systems (1) overlook tool-return readiness in service evaluation, wasting limited resources, and (2) use heuristic strategies to decide admission and placement on GPUs, resulting in suboptimal scheduling decisions. To bridge this gap, we present TAO, a tool-aware joint scheduling framework for LLM-based agentic serving. Its value interface combines near-term too},
url = {https://huggingface.co/papers/2610.37230},
keywords = {LLM-based agentic serving, tool-aware scheduling, tool-return readiness, node-specific cache credit, time-aware decay, token footprint, dynamic programming, admission and placement, decode throughput, time to first token (TTFT), huggingface daily},
eprint = {2610.37230},
archiveprefix = {arXiv},
}
{}