Home
An enterprise-grade large language model (LLM) designed for secure use in private environments.
Values of Takane
High Precision Japanese language processing
Takane delivers world-class Japanese language capabilities. Based on the highly accurate and customizable Cohere Command series, Fujitsu has enhanced its performance through proprietary Japanese language training.
Secure deployment in private environments
Ideal for customers requiring advanced security—such as financial institutions, government agencies, and R&D departments—Takane can be deployed in private environments. Sensitive business data can be safely leveraged with generative AI.
Customise to meet your requirements
Takane can be finely tuned to align with your organization’s specific needs. Unlike generic LLMs, it supports advanced tuning using your trusted business data, enabling customization for your workflows. Additionally, world-class Retrieval-Augmented Generation (RAG) technology ensures optimal performance for your operations.
High performance with minimal infrastructure resources
When running on NVIDIA H100, Takane operates with only one-eighth of the resources compared to GPT-4o, delivering maximum performance with minimal infrastructure investment.（Source : https://cohere.com/blog/command-a）
Use cases
Takane is ideal for the industries handling highly confidential data—such as finance, manufacturing (design and development), and national security—where public generative AI services pose data leakage risks.
Use Case 1 – Streamlining Support Operations
-
Utilize confidential data such as troubleshooting history, manuals, design documents, and source code to provide rapid responses to the customers.
-
When FAQs are inadequate or difficult to find a solution, a chatbot can help the customers or operators to find a proper solution.
Use Case 2 – Visualizing patient information over time
Using Takane to extract and visualize patient data such as severity classification, disease types, and vital signs from large-scale medical records like electronic health records.
Even if a record states, “symptoms appeared the next morning,” Takane can infer the actual date based on context and hospitalization date, enabling chronological output of patient data.
Product offering
Takane is available through Fujitsu Generative AI for Cohere. In addition to large-scale models, we also offer small and high-speed models that minimize infrastructure requirements.
Product lineup
Usage options
Takane can be used in three different forms. Select the option which works best for you.
Three generative AI solutions for handling highly confidential data
Co-development
Takane is co-developed with Cohere Inc. By combining the strengths of both the companies, Takane enables easy and smooth handling of diverse and large-scale enterprise data, delivering new value through generative AI across various business scenarios.
