Cost-Optimal Foundation Model Deployment Portfolio for Transportation Management
Xi Cheng, Ke Liu, Siyuan Feng, Jane Lin, H. Oliver Gao
Why It Matters
What makes this one worth your time
This research provides a practical framework for optimizing the deployment of AI models in transportation management, potentially reducing costs and improving efficiency.
A cost-effective strategy for deploying foundation models in transportation management.
Summary
The paper addresses the deployment of foundation models for transportation management tasks by formulating the Foundation Model Deployment Portfolio (FMDP) problem, which aims to minimize total cost of ownership while meeting quality, latency, and safety constraints. The problem is proven NP-hard, and a polynomial-time greedy heuristic is proposed. A case study demonstrates significant cost savings by using a mixed portfolio of open-source and closed APIs.
Key contributions
- Formulation of the Foundation Model Deployment Portfolio (FMDP) problem.
- Proof of NP-hardness for the FMDP problem.
- Development of a polynomial-time greedy heuristic for solving the FMDP problem.
Notable insights
- The FMDP problem is proven NP-hard, indicating its computational complexity.
- A polynomial-time greedy heuristic is proposed as a solution, suggesting a practical approach to a complex problem.
Possible limitations
- Not stated in the abstract
Abstract
arXiv:2607.13239v1 Announce Type: new Abstract: Foundation models, including large language models (LLMs) and vision-language models (VLMs), are increasingly used for transportation management center (TMC) tasks such as anomaly detection, incident reporting, and traveler information. Deploying multiple such models across TMC functions raises a portfolio question: which model should serve each function, in which deployment mode, and under what shared hardware budget? We formulate this as the Foundation Model Deployment Portfolio (FMDP) problem, a mixed-integer program minimizing total cost of ownership (TCO) subject to per-function quality, latency, and safety constraints over shared GPU capacity. We prove the problem NP-hard by reduction from the 0-1 knapsack problem and propose a polynomial-time greedy heuristic. In an illustrative case study with five TMC functions and 19 candidate (model, mode) pairs, FMDP identifies a mixed portfolio costing $34/mo (97% below the cheapest feasible all-closed-API baseline) by routing four functions to open-source APIs and the one function whose quality floor no open-source model meets to a closed API. Break-even analysis shows that on-premise GPU investment becomes reasonable only above approximately 309 vision queries/hour or if API prices double.