---
title: "PromptBridge: Cross-Model Prompt Transfer for Large Language Models"
canonical_url: "https://www.modelscope.cn/papers/218071"
md_url: "https://www.modelscope.cn/papers/218071.md"
arxiv_id: 2512.01420
published: 2025-12-01
last_updated: 2025-12-01
authors:
  - "Yaxuan Wang"
  - "Quan Liu"
  - "Zhenting Wang"
  - "Zichao Li"
  - "Wei Wei"
  - "Yang Liu"
  - "Yujia Bao"
model_name: PromptBridge
model_developer: "University of California、Santa Cruz、Center for Advanced AI、Accenture"
domain:
  - "自然语言处理"
  - "人工智能"
  - "提示工程"
  - "大语言模型"
  - "智能体系统"
type:
  - "自然语言处理"
  - "人工智能"
  - "提示工程"
  - "大语言模型"
  - "智能体系统"
  - "Computation and Language"
  - "Artificial Intelligence"
arxiv_url: "https://arxiv.org/abs/2512.01420"
pdf_url: "https://arxiv.org/pdf/2512.01420"
code_link: "https://github.com/supergirl-os/PromptBridge"
---

# PromptBridge: Cross-Model Prompt Transfer for Large Language Models

> Large language models (LLMs) underpin applications in code generation, mathematical reasoning, and agent-based workflows. In practice, systems access LLMs via commercial APIs or open-source deployments, and the model landscape (e.g., GPT, Claude, Llama)…

「PromptBridge: Cross-Model Prompt Transfer for Large Language Models」是 ModelScope 魔搭社区收录的论文，arXiv 2512.01420，作者为 Yaxuan Wang, Quan Liu, Zhenting Wang et al.，发表于 2025-12-01，属于 自然语言处理、人工智能、提示工程 领域。

- **ArXiv**: 2512.01420
- **Published**: 2025-12-01
- **Authors**: Yaxuan Wang, Quan Liu, Zhenting Wang, Zichao Li, Wei Wei, Yang Liu, Yujia Bao
- **Model**: PromptBridge
- **Developer**: University of California、Santa Cruz、Center for Advanced AI、Accenture
- **Domain**: 自然语言处理, 人工智能, 提示工程, 大语言模型, 智能体系统
- **ArXiv URL**: https://arxiv.org/abs/2512.01420
- **PDF**: https://arxiv.org/pdf/2512.01420
- **Code**: https://github.com/supergirl-os/PromptBridge

Source: https://www.modelscope.cn/papers/218071

---

> PromptBridge：打通大模型换代鸿沟的零样本提示迁移利器

## 摘要

本文针对大语言模型（LLM）在实际应用中频繁更换模型所带来的‘模型漂移’问题，提出了一种无需训练的跨模型提示迁移框架PromptBridge。由于不同LLM在训练数据、对齐策略和接口设计上的差异，为某一模型精心优化的提示（prompt）在迁移到另一模型时常导致性能显著下降，这一现象被称为‘Model Drifting’。为量化该问题，作者通过大量实验验证了跨模型提示不兼容性的普遍性和严重性。为此，本文提出了PromptBridge框架：首先利用提出的Model-Adaptive Reflective Prompt Evolution（MAP-RPE）方法，在少量对齐任务上分别生成源模型和目标模型的最优提示；然后基于这些配对提示学习一个通用的提示映射函数；在测试阶段，对于新任务的源模型提示，该函数可直接生成适配目标模型的高质量提示，实现零样本迁移。实验涵盖单智能体与多智能体场景，在SWE-BENCH和TERMINAL-BENCH等基准上，相比直接迁移，PromptBridge分别带来27.39%和39.44%的性能提升，有效降低了模型升级或替换时的提示重构成本。

## Abstract

Large language models (LLMs) underpin applications in code generation, mathematical reasoning, and agent-based workflows. In practice, systems access LLMs via commercial APIs or open-source deployments, and the model landscape (e.g., GPT, Claude, Llama) evolves rapidly. This rapid evolution forces frequent model switches driven by capability, cost, deployment constraints, and privacy. Yet prompts are highly model-sensitive: reusing a prompt engineered for one model on another often yields substantially worse performance than a prompt optimized for the target model. We term this phenomenon Model Drifting. Through extensive empirical analysis across diverse LLM configurations, we show that model drifting is both common and severe. To address this challenge, we introduce PromptBridge, a training-free framework that preserves prompt effectiveness under model switches, enabling cross-model prompt transfer without costly per-task or per-model re-optimization. PromptBridge requires only a small set of alignment tasks for calibration. It first applies Model-Adaptive Reflective Prompt Evolution (MAP-RPE) to obtain task- and model-specific optimal prompts via iterative reflective refinement and quantitative evaluation. Using the resulting calibrated prompt pairs for the source and target models, PromptBridge learns a cross-model prompt mapping. At test time, i.e., for an unseen task, given a source-model prompt, this mapping directly produces an optimized prompt for the target model. Experiments in single-agent and multi-agent settings show that PromptBridge consistently improves downstream accuracy while reducing migration effort. The code will be available soon.