Dynamic Model Routing and Cascading for Efficient LLM Inference: A Survey
Yasmin Moslem [email protected], John D. Kelleher [email protected] · 2026
Read Dynamic Model Routing and Cascading for Efficient LLM Inference: A Survey by Yasmin Moslem [email protected], John D. Kelleher [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-04445.txt.
Description
Moslem and Kelleher survey dynamic routing and cascading for language model inference. Static use ignores query hardness and domain, wasting performance and money. The survey maps kinds from difficulty and preference to clustering, uncertainty, reinforcement learning, multimodality, and cascading, with methods and trade offs for each. A framework sorts systems by decision timing, information, and computation, showing real systems blend kinds. Good routing can beat the strongest single model by exploiting specialist strengths. The best pick thus depends on the job at hand.
Catalog record
| CREATOR | Yasmin Moslem [email protected], John D. Kelleher [email protected] |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50828 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.04445