Problem
Research questions and friction points this paper is trying to address.
Financial Agents
Reliability
Long-horizon Execution
Auditable Evidence
Agentic Foundation Models
Mint-Agent: Introducing Finance-Native Agentic Foundation Models
🤖 AI Summary
This study addresses the reliability deficits and lack of long-horizon auditability in financial agents by proposing finance-native agents built upon three pillars: a data engine, an interaction framework, and hybrid training. The approach integrates multi-teacher distillation to unify reasoning and execution capabilities, optimized via supervised fine-tuning, key-step on-policy distillation, reinforcement learning with verifiable rewards, and model merging. Experimental results demonstrate that Mint-AG achieves 98.33% accuracy on RFC-Bench, surpassing GPT-5.6, while Mint-CU outperforms baselines by 22.83% on FinSearchComp. These findings establish a breakthrough in general-purpose financial modeling, simultaneously achieving precise operational execution and long-horizon auditability for rigorous financial research.
📝 Abstract
Financial agents must do more than recall domain knowledge: they must be both reliable, executing precise operations over grounded evidence, and executive, sustaining long-horizon research whose conclusions remain auditable. We present Mint-Agent, a family of finance-native agentic models designed around these two scales of financial intelligence. Mint-Agent is built upon three pillars: data, harness, and algorithm. Our data engine constructs clean, specialized tasks for atomic financial capabilities and long-horizon agentic execution from real-world financial sources. MintHarness enables stable interaction with open-ended environments and maintains auditable evidence trails across extended research trajectories. Our training recipe combines SFT, critical-step OPD, and RLVR to develop separate financial reasoning and agentic execution experts, which are then unified through model merging and multi-teacher on-policy distillation into compact, general-purpose financial agents. This pipeline yields two flagship models, Mint-Cu (9B) and Mint-Ag (27B). Across professional financial benchmarks, our models demonstrate two defining strengths: (1) Reliability: Mint-Ag achieves 98.33% on RFC-Bench, surpassing GPT-5.6-Sol and Claude-Opus-4.8 by 3.66 and 3.00 points; and (2) Executability: Mint-Cu reaches 69.86% on FinSearchComp T2, outperforming Agents-A1-35B and Nex-N2-mini by 22.83 and 12.78 points, while Mint-Ag achieves 76.00% and 60.49% on FinanceAgentBench v1.1 and v2, respectively. These results establish a path toward trustworthy financial intelligence in which domain expertise, long-horizon execution, and auditable evidence are jointly engineered as a unified foundation for frontier agentic models.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Finance-Native Agentic Model
Auditable Evidence Trail
Multi-Teacher On-Policy Distillation
Critical-Step OPD
Long-Horizon Execution
🔎 Similar Papers
