通过动态路由让大模型更便宜又更好，自动选最合适的模型处理每条请求。
Beyond GPT-5: Making LLMs Cheaper and Better via Performance-Efficiency Optimized Routing
- 根据查询复杂度动态分配不同能力的模型，统一优化性能与效率。
- 在6个基准上平均准确率比最强单模型高7%，成本低27%仍达同等效果。
- 适合追求性价比的AI系统部署，尤其适合资源受限场景。
平衡性能与效率是大语言模型发展的核心挑战。GPT-5采用推理时路由机制，动态将请求分配给高效或高容量模型。本文提出Avengers-Pro，一种集成多能力、多效率模型的推理时路由框架，可统一应对所有性能-效率权衡问题。该方法对输入查询进行嵌入与聚类，并基于性能-效率得分将其路由至最优模型。在6个挑战性基准和8个主流模型（包括GPT-5-medium、Gemini-2.5-pro、Claude-opus-4.1）上，Avengers-Pro实现业界最优表现：通过调节权衡参数，平均准确率超越最强单模型（GPT-5-medium）达+7%；在27%更低成本下达到相同平均准确率，或以63%成本达成约90%性能。更重要的是，其构建了帕累托前沿，在任意成本下均获得最高准确率，或在任意准确率下实现最低成本。代码已开源：https://github.com/ZhangYiqun018/AvengersPro。
原文摘要 · Abstract (English)
Balancing performance and efficiency is a central challenge in large language model (LLM) advancement. GPT-5 addresses this with test-time routing, dynamically assigning queries to either an efficient or a high-capacity model during inference. In this work, we present Avengers-Pro, a test-time routing framework that ensembles LLMs of varying capacities and efficiencies, providing a unified solution for all performance-efficiency tradeoffs. The Avengers-Pro embeds and clusters incoming queries, then routes each to the most suitable model based on a performance-efficiency score. Across 6 challenging benchmarks and 8 leading models -- including GPT-5-medium, Gemini-2.5-pro, and Claude-opus-4.1 -- Avengers-Pro achieves state-of-the-art results: by varying a performance-efficiency trade-off parameter, it can surpass the strongest single model (GPT-5-medium) by +7% in average accuracy. Moreover, it can match the average accuracy of the strongest single model at 27% lower cost, and reach ~90% of that performance at 63% lower cost. Last but not least, it achieves a Pareto frontier, consistently yielding the highest accuracy for any given cost, and the lowest cost for any given accuracy, among all single models. Code is available at https://github.com/ZhangYiqun018/AvengersPro.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。