Abstract
Foundation models, including large language models (LLMs) and vision-language models (VLMs), are increasingly utilized in transportation management center (TMC) tasks such as anomaly detection, incident reporting, and traveler information. The deployment of multiple models 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 aimed at minimizing total cost of ownership (TCO) while adhering to per-function quality, latency, and safety constraints based on shared GPU capacity. We prove the problem is NP-hard by reduction from the 0-1 knapsack problem and propose a polynomial-time greedy heuristic. In a case study involving five TMC functions and 19 candidate (model, mode) pairs, FMDP identifies a mixed portfolio costing $34/month (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 indicates that on-premise GPU investment becomes reasonable only when exceeding approximately 309 vision queries/hour or if API prices double.
Blogger's Review: This research systematically addresses the deployment issue of foundation models in transportation management using mixed-integer programming, showcasing a cost-benefit analysis of open-source versus closed APIs. The specific case study provides practical insights that contribute to the ongoing intelligent transformation in the transportation sector.