OLAF: Next-Generation LLM for Financial Content Creation

Finola

・

Kickstart Your Growth with Finola

Try Now

About

Products

Career

Blog

Contact Us

Finola

・

Kickstart Your Growth with Finola

Try Now

Model

The Powerhouse of OneLineAI’s On-Prem LLM

Behind both OLA and Finola is OneLineAI’s proprietary LLM—designed to run securely on your premises or in the cloud, ensuring high performance, minimized data transfer, and privacy compliance for the most sensitive financial information.

Features

Key Features

Inside Secure, Enterprise-Grade LLMs

Financial Reasoning Model

Advanced reasoning for complex financial tasks

Financial Reasoning Model

Advanced reasoning for complex financial tasks

Financial Reasoning Model

Advanced reasoning for complex financial tasks

Test-Time Scaling

Scale resources at inference for near-large-model performance

Test-Time Scaling

Scale resources at inference for near-large-model performance

Test-Time Scaling

Scale resources at inference for near-large-model performance

Switch Thinking & Non-Thinking

Toggle between deep reasoning and quick responses as needed

Switch Thinking & Non-Thinking

Toggle between deep reasoning and quick responses as needed

Switch Thinking & Non-Thinking

Toggle between deep reasoning and quick responses as needed

32K Context Length

Optimized for large-scale financial document processing

32K Context Length

Optimized for large-scale financial document processing

32K Context Length

Optimized for large-scale financial document processing

Safety & Refusal Mechanisms

Minimizes hallucinations and improves reliability

Safety & Refusal Mechanisms

Minimizes hallucinations and improves reliability

Safety & Refusal Mechanisms

Minimizes hallucinations and improves reliability

14B & 1.5B Models

Choose between high-performance or lightweight options

14B & 1.5B Models

Choose between high-performance or lightweight options

14B & 1.5B Models

Choose between high-performance or lightweight options

Thinking Mode

Switch Thinking &
Non-Thinking Mode

OLAF2 swiftly switches between Thinking Mode and Non-Thinking Mode, providing deep analysis for complex tasks like financial report generation, while delivering quick answers for simple queries or information retrieval.

Thinking Mode

Ideal for multi-step math calculations, detailed financial analyses, and scenario forecasting where layered reasoning is essential

Non-Thinking Mode

Perfect for quick Q&A, rapid information lookup, and concise summaries that require instant responses

Thinking Mode

Ideal for multi-step math calculations, detailed financial analyses, and scenario forecasting where layered reasoning is essential

Non-Thinking Mode

Perfect for quick Q&A, rapid information lookup, and concise summaries that require instant responses

Thinking Mode

Ideal for multi-step math calculations, detailed financial analyses, and scenario forecasting where layered reasoning is essential

Non-Thinking Mode

Perfect for quick Q&A, rapid information lookup, and concise summaries that require instant responses

Scaling

Test-Time Scaling

OLAF2 scales computing resources at inference time, enabling smaller models to achieve performance close to that of larger models. Even in constrained in-house server settings, it efficiently handles financial data, and additional resources can be deployed when deeper analysis is required.

Performance

Model Performance

OLAF2 has demonstrated exceptional scores across various benchmarks.

HRM8K

(Math Reasoning)

OLAF2-14B (Thinking Mode)

45.8 Score

OLAF2-14B (Thinking Mode)

45.8 Score

OLAF2-14B (Thinking Mode)

45.8 Score

Llama-3.1-70B

45.6 Score

Llama-3.1-70B

45.6 Score

Llama-3.1-70B

45.6 Score

Qwen2.5-32B

44.4 Score

Qwen2.5-32B

44.4 Score

Qwen2.5-32B

44.4 Score

KMMLU

(Korean Knowledge Evaluation)

OLAF2-14B

54.21 Score

OLAF2-14B

54.21 Score

OLAF2-14B

54.21 Score

EXAONE-3.5-32B

47.63 Score

EXAONE-3.5-32B

47.63 Score

EXAONE-3.5-32B

47.63 Score

Llama-3.1-70B

51.83 Score

Llama-3.1-70B

51.83 Score

Llama-3.1-70B

51.83 Score

HRM8K-Small Models

(Math Reasoning)

OLAF2-1.5B (Thinking Mode)

38.0 Score

OLAF2-1.5B (Thinking Mode)

38.0 Score

OLAF2-1.5B (Thinking Mode)

38.0 Score

EXAONE-3.5-2.4B-Instruct

17.6 Score

EXAONE-3.5-2.4B-Instruct

17.6 Score

EXAONE-3.5-2.4B-Instruct

17.6 Score

LogicKor

(Korean Instruction-Following Benchmark)

OLAF2-14B

8.51 Score

OLAF2-14B

8.51 Score

OLAF2-14B

8.51 Score

Llama-3.1-70B

7.78 Score

Llama-3.1-70B

7.78 Score

Llama-3.1-70B

7.78 Score

OLAF2: Elevate Your Financial Intelligence

The finance‑focused AI that delivers precise insights at speed. Choose from 14B or 1.5B models to match your business needs.

Finance-Focused AI

OLAF2 is a specialized AI solution designed to handle everything from complex math and logic tasks to simple information retrieval in the financial domain.

Finance-Focused AI

OLAF2 is a specialized AI solution designed to handle everything from complex math and logic tasks to simple information retrieval in the financial domain.

Finance-Focused AI

OLAF2 is a specialized AI solution designed to handle everything from complex math and logic tasks to simple information retrieval in the financial domain.

14B & 1.5B Versions

Both models (14B and 1.5B) offer fast, stable performance and can be tailored to different business scales and requirements.

14B & 1.5B Versions

Both models (14B and 1.5B) offer fast, stable performance and can be tailored to different business scales and requirements.

14B & 1.5B Versions

Both models (14B and 1.5B) offer fast, stable performance and can be tailored to different business scales and requirements.

Test-Time Scaling & Thinking Mode

Smaller models can approach large-model performance through resource scaling, and users can switch between deep reasoning (Thinking Mode) and quick responses (Non-Thinking Mode) based on task complexity.

Test-Time Scaling & Thinking Mode

Implementation Benefits

By integrating OLAF2, you can unlock new opportunities in financial data analysis—whether it’s rapid and accurate report generation or in-depth financial forecasting—enabling a more efficient workflow.