Fin-R1: A 7B-Parameter Financial Large Language Model Trained with Reinforcement Learning, Outperforming Industry Giants Based on Qwen2.5-7B

AIbase基地

Published inAI News · 4 min read · Mar 24, 2025

A powerful new contender has emerged in the FinTech arena. The Fin-R1 model, jointly developed by Professor Zhang Liwen's team at the School of Statistics and Data Science, Shanghai University of Finance and Economics (SUFE-AIFLM-Lab), and Caiyue Xingchen, has been officially open-sourced, garnering significant industry attention due to its impressive performance.

This financial-specific large language model, based on Qwen2.5-7B and trained using reinforcement learning, achieves leading performance across multiple financial benchmark tests.

Investment, Finance, Money

Remarkably, Fin-R1, with only 7B parameters, surpasses most competitors of similar size, and even those with tens of times more parameters. It excels in crucial tasks such as Financial Table Reasoning (FinQA) and Conversational Financial QA (ConvFinQA), demonstrating exceptional understanding of the financial domain.

Tailored for reasoning and analytical tasks in core financial business scenarios, Fin-R1 boasts an extensive range of capabilities. It can proficiently write financial code, build pricing models and risk assessment scripts; accurately perform quantitative analysis and report calculations; fluently generate English financial models and professional reports; provide financial security and compliance analysis; implement anti-fraud measures, default prediction, and other intelligent risk control functions; and even conduct ESG (Environmental, Social, and Governance) sustainability analysis.

Fin-R1's development leveraged advanced techniques. The research team built the model architecture based on Qwen2.5-7B-Instruct and utilized the DeepSeek-R1 framework for "data distillation" and "dual-wheel quality screening." A training method combining supervised fine-tuning (SFT) with high-quality chain-of-thought data and reinforcement learning (RL) successfully created this AI assistant for the financial industry.

Notably, Fin-R1 supports both Chinese and English environments, enabling financial modeling, report generation, and conversational interaction in both languages, showcasing strong cross-lingual capabilities. The release of this open-source model will provide strong support for the digital transformation of the financial industry and promises to become a valuable tool for financial analysts, risk management experts, and investment advisors.

Tencent's HunYuan-T1 Reasoning Model Matches OpenAI's Top Performance in Benchmark Tests

Tencent recently announced its latest large language model, HunYuan-T1, claiming its reasoning capabilities rival OpenAI's best reasoning systems. Tencent reports that HunYuan-T1's development heavily relied on reinforcement learning, with 96.7% of post-training computing power dedicated to enhancing its logical reasoning and alignment with human preferences. In various benchmark tests, HunYuan-T1 demonstrated strong performance. On the MMLU-PRO evaluation, testing knowledge across 14 academic subjects, the model achieved a score of 87.2.

360 ZhiNao Team Successfully Replicates DeepSeek Reinforcement Learning Results, Releases Open-Source Model Light-R1-14B-DS

Recently, the 360 ZhiNao team announced the successful replication of DeepSeek's reinforcement learning results and the official release of the open-source reasoning model Light-R1-14B-DS. This model surpasses DeepSeek-R1-Distill-Llama-70B and DeepSeek-R1-Distill-Qwen-32B in performance, becoming the industry's first 14B parameter model to achieve reinforcement learning effects. It significantly improves mathematical reasoning capabilities, outperforming most 32B-parameter models.

Ant Group's Medical Large Language Model Wins Double Championship in MedBench Evaluation, Ushering in a New Era for Medical AI

Recently, MedBench, a well-known domestic medical large language model evaluation platform, released its latest rankings. Ant Group's medical team's self-developed medical large language model achieved outstanding performance, winning first place in both the evaluation and self-test rankings with high scores of 97.5 and 98.2 respectively, attracting widespread attention from the industry. The success of Ant Group's medical large language model is inseparable from the team's continuous efforts in researching and developing medical reasoning models. The team recently adopted reinforcement learning technology to create a new generation of medical reasoning models. This innovation has enabled the model to...

Alibaba's Tongyi Open-Sources R1-Omni Model for Enhanced Multimodal Emotion Recognition

On March 11th, the Tongyi Lab team announced the open-sourcing of the R1-Omni model, marking a significant advancement in multimodal model development. This model integrates reinforcement learning with verifiable reward (RLVR) methods, focusing on improving reasoning capabilities and generalization performance in multimodal emotion recognition tasks. R1-Omni's training is divided into two stages. In the cold-start phase, the team fine-tuned the model using a combined dataset containing 580 video clips sourced from Explainable Multimodal Emotion...