Explainable Model Routing for Agentic Workflows

2026-04-03

AI Summary

This work addresses the lack of interpretability in model routing within existing agent workflows, where current approaches prioritize performance optimization without transparently accounting for trade-offs between cost and capability—making it difficult for developers to distinguish between efficient scheduling and failures caused by budget constraints. To bridge this gap, we propose Topaz, a novel framework that introduces interpretability and auditability into model routing. Topaz achieves this through skill-oriented model profiling, a traceable routing algorithm that jointly optimizes multiple objectives under budget constraints, and natural language explanation generation tailored for developers. By rendering routing decisions transparent and controllable, Topaz significantly enhances developers’ understanding, trust, and ability to intervene, enabling efficient iterative tuning of cost–quality trade-offs.

Abstract

Modern agentic workflows decompose complex tasks into specialized subtasks and route them to diverse models to minimize cost without sacrificing quality. However, current routing architectures focus exclusively on performance optimization, leaving underlying trade-offs between model capability and cost unrecorded. Without clear rationale, developers cannot distinguish between intelligent efficiency -- using specialized models for appropriate tasks -- and latent failures caused by budget-driven model selection. We present Topaz, a framework that introduces formal auditability to agentic routing. Topaz replaces silent model assignments with an inherently interpretable router that incorporates three components: (i) skill-based profiling that synthesizes performance across diverse benchmarks into granular capability profiles (ii) fully traceable routing algorithms that utilize budget-based and multi-objective optimization to produce clear traces of how skill-match scores were weighed against costs, and (iii) developer-facing explanations that translate these traces into natural language, allowing users to audit system logic and iteratively tune the cost-quality tradeoff. By making routing decisions interpretable, Topaz enables users to understand, trust, and meaningfully steer routed agentic systems.

Problem

• model routing
• agentic workflows
• explainability
• cost-quality tradeoff
• auditability

Innovation

• explainable AI
• model routing
• agentic workflows
• interpretable router
• cost-quality tradeoff