Autellix: An Efficient Serving Engine for LLM Agents as General Programs
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929720500682752 |
|---|---|
| author | Luo, Michael Shi, Xiaoxiang Cai, Colin Zhang, Tianjun Wong, Justin Wang, Yichuan Wang, Chi Huang, Yanping Chen, Zhifeng Gonzalez, Joseph E. Stoica, Ion |
| author_facet | Luo, Michael Shi, Xiaoxiang Cai, Colin Zhang, Tianjun Wong, Justin Wang, Yichuan Wang, Chi Huang, Yanping Chen, Zhifeng Gonzalez, Joseph E. Stoica, Ion |
| contents | Large language model (LLM) applications are evolving beyond simple chatbots into dynamic, general-purpose agentic programs, which scale LLM calls and output tokens to help AI agents reason, explore, and solve complex tasks. However, existing LLM serving systems ignore dependencies between programs and calls, missing significant opportunities for optimization. Our analysis reveals that programs submitted to LLM serving engines experience long cumulative wait times, primarily due to head-of-line blocking at both the individual LLM request and the program. To address this, we introduce Autellix, an LLM serving system that treats programs as first-class citizens to minimize their end-to-end latencies. Autellix intercepts LLM calls submitted by programs, enriching schedulers with program-level context. We propose two scheduling algorithms-for single-threaded and distributed programs-that preempt and prioritize LLM calls based on their programs' previously completed calls. Our evaluation demonstrates that across diverse LLMs and agentic workloads, Autellix improves throughput of programs by 4-15x at the same latency compared to state-of-the-art systems, such as vLLM. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_13965 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Autellix: An Efficient Serving Engine for LLM Agents as General Programs Luo, Michael Shi, Xiaoxiang Cai, Colin Zhang, Tianjun Wong, Justin Wang, Yichuan Wang, Chi Huang, Yanping Chen, Zhifeng Gonzalez, Joseph E. Stoica, Ion Machine Learning Artificial Intelligence Distributed, Parallel, and Cluster Computing Large language model (LLM) applications are evolving beyond simple chatbots into dynamic, general-purpose agentic programs, which scale LLM calls and output tokens to help AI agents reason, explore, and solve complex tasks. However, existing LLM serving systems ignore dependencies between programs and calls, missing significant opportunities for optimization. Our analysis reveals that programs submitted to LLM serving engines experience long cumulative wait times, primarily due to head-of-line blocking at both the individual LLM request and the program. To address this, we introduce Autellix, an LLM serving system that treats programs as first-class citizens to minimize their end-to-end latencies. Autellix intercepts LLM calls submitted by programs, enriching schedulers with program-level context. We propose two scheduling algorithms-for single-threaded and distributed programs-that preempt and prioritize LLM calls based on their programs' previously completed calls. Our evaluation demonstrates that across diverse LLMs and agentic workloads, Autellix improves throughput of programs by 4-15x at the same latency compared to state-of-the-art systems, such as vLLM. |
| title | Autellix: An Efficient Serving Engine for LLM Agents as General Programs |
| topic | Machine Learning Artificial Intelligence Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2502.13965 |