Instructions to use toolevalxm/FinanceGPT-TradingAssist-BestModel with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use toolevalxm/FinanceGPT-TradingAssist-BestModel with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="toolevalxm/FinanceGPT-TradingAssist-BestModel")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("toolevalxm/FinanceGPT-TradingAssist-BestModel")
model = AutoModelForCausalLM.from_pretrained("toolevalxm/FinanceGPT-TradingAssist-BestModel")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use toolevalxm/FinanceGPT-TradingAssist-BestModel with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "toolevalxm/FinanceGPT-TradingAssist-BestModel"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "toolevalxm/FinanceGPT-TradingAssist-BestModel",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/toolevalxm/FinanceGPT-TradingAssist-BestModel

SGLang

How to use toolevalxm/FinanceGPT-TradingAssist-BestModel with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "toolevalxm/FinanceGPT-TradingAssist-BestModel" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "toolevalxm/FinanceGPT-TradingAssist-BestModel",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "toolevalxm/FinanceGPT-TradingAssist-BestModel" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "toolevalxm/FinanceGPT-TradingAssist-BestModel",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use toolevalxm/FinanceGPT-TradingAssist-BestModel with Docker Model Runner:
```
docker model run hf.co/toolevalxm/FinanceGPT-TradingAssist-BestModel
```

FinanceGPT-TradingAssist

1. Introduction

The FinanceGPT-TradingAssist has undergone a significant version upgrade. In the latest update, FinanceGPT has significantly improved its financial analysis and market prediction capabilities by leveraging increased computational resources and introducing algorithmic optimization mechanisms during post-training. The model has demonstrated outstanding performance across various financial benchmark evaluations, including risk assessment, fraud detection, and portfolio optimization. Its overall performance is now approaching that of other leading financial AI models.

Compared to the previous version, the upgraded model shows significant improvements in handling complex financial analysis tasks. For instance, in the FinBench 2025 test, the model's accuracy has increased from 68% in the previous version to 89.2% in the current version. This advancement stems from enhanced financial reasoning depth during the analysis process: in the FinBench test set, the previous model used an average of 8K tokens per analysis, whereas the new version averages 18K tokens per analysis.

Beyond its improved financial reasoning capabilities, this version also offers a reduced hallucination rate and enhanced support for real-time market data integration.

2. Evaluation Results

Comprehensive Benchmark Results

	Benchmark	BaselineA	BaselineB	BaselineA-v2	FinanceGPT
Risk Analysis	Risk Assessment	0.620	0.645	0.651	0.836
	Fraud Detection	0.789	0.812	0.820	0.875
	Anomaly Detection	0.716	0.732	0.745	0.820
Market Intelligence	Market Prediction	0.571	0.595	0.610	0.706
	Price Forecasting	0.582	0.609	0.621	0.718
	Financial Sentiment	0.803	0.821	0.830	0.870
	Financial QA	0.677	0.691	0.700	0.774
Portfolio Management	Portfolio Optimization	0.615	0.641	0.660	0.751
	Investment Recommendation	0.588	0.609	0.621	0.720
	Credit Scoring	0.721	0.745	0.759	0.821
	Loan Approval	0.745	0.765	0.780	0.835
Compliance & Reporting	Regulatory Compliance	0.782	0.809	0.821	0.864
	Tax Analysis	0.651	0.678	0.690	0.774
	Report Generation	0.733	0.759	0.771	0.821
	Financial Summarization	0.718	0.741	0.755	0.799

Overall Performance Summary

The FinanceGPT-TradingAssist demonstrates strong performance across all evaluated financial benchmark categories, with particularly notable results in risk analysis and portfolio management tasks.

3. Trading Dashboard & API Platform

We offer a trading dashboard and API for you to integrate FinanceGPT into your financial applications. Please check our official website for more details.

4. How to Run Locally

Please refer to our code repository for more information about running FinanceGPT locally.

Compared to previous versions, the usage recommendations for FinanceGPT have the following changes:

Market data context is now supported.
It is not required to add special tokens at the beginning of the output to force the model into a specific analysis pattern.

The model architecture of FinanceGPT-Lite is identical to its base model, but it shares the same tokenizer configuration as the main FinanceGPT. This model can be run in the same manner as its base model.

System Prompt

We recommend using the following system prompt with market context.

You are FinanceGPT, a professional financial AI assistant.
Current market date is {current date}.
Market status: {market_status}.

For example,

You are FinanceGPT, a professional financial AI assistant.
Current market date is May 28, 2025, Monday.
Market status: NYSE Open, NASDAQ Open.

Temperature

We recommend setting the temperature parameter $T_{model}$ to 0.3 for financial analysis tasks (lower for more consistent outputs).

Prompts for Financial Data Analysis

For financial report analysis, please follow the template to create prompts, where {ticker}, {report_content} and {analysis_request} are arguments.

finance_template = \
"""[Ticker Symbol]: {ticker}
[Financial Report Begin]
{report_content}
[Financial Report End]
{analysis_request}"""

For market analysis with real-time data, we recommend the following prompt template where {market_data}, {cur_date}, and {analysis_request} are arguments.

market_analysis_template = \
'''# Current Market Data:
{market_data}
In the market data provided, each entry is formatted as [TICKER: PRICE | CHANGE | VOLUME], where TICKER represents the stock symbol. Please cite specific data points when making recommendations. Use the citation format [ticker:SYMBOL] in the corresponding part of your analysis.
When responding, please keep the following points in mind:
- Today is {cur_date}.
- Not all market data is relevant to the analysis request. Focus on sectors and tickers mentioned.
- For portfolio recommendations, limit suggestions to 10 key positions with clear rationale.
- For risk assessments, quantify potential downside scenarios.
- Always include appropriate risk disclaimers for investment advice.
# The analysis request is:
{analysis_request}'''

5. License

This code repository is licensed under the Apache 2.0 License. The use of FinanceGPT models is also subject to the Apache 2.0 License. The model series supports commercial use with proper compliance.

6. Contact

If you have any questions, please raise an issue on our GitHub repository or contact us at support@financegpt.ai.

Downloads last month: 4