TOKYO – December 2, 2024 – Preferred Networks, Inc. (PFN) today launched PLaMo™ Prime, its new flagship model in the PLaMo line of large language models (LLMs) developed in Japan by its subsidiary Preferred Elements, Inc. (PFE). PLaMo Prime is now commercially available via PLaMo API and is accessible for free via the PLaMo Chat assistant during the initial trial period.

PLaMo website (Japanese): https://plamo.preferredai.jp/

Based on the user feedback from the recent PLaMo beta trial, PLaMo Prime has been trained with a high-quality data set and additionally trained for Japanese-language tasks, resulting in its world-class Japanese-language performance while keeping the model size compact. PLaMo Prime can now perform better in a range of natural language tasks including text generation, summarization, translation and text analysis. 

For PLaMo Prime’s latest technical improvements, please read the PFN Tech Blog:

PLaMo Prime release & feature improvement – PFN Tech Blog

How to use PLaMo Prime

1. PLaMo API

PLaMo API users can easily incorporate PLaMo in systems and applications of their choice. Since PLaMo API’s interface is compatible with OpenAI API, users can keep existing codes and still use LLM libraries such as openai-python and LangChain.

Pricing: PLaMo API is available at affordable rates compared with other LLMs with similar Japanese-language performance. During the launch campaign period, 1,000 yen of credits will be given to new users.

・300 yen / 1M input tokens
・1,000 yen / 1M input tokens

2. PLaMo Chat

PLaMo Chat is an AI assistant designed for the general audience who wish to try PLaMo Prime. PLaMo Chat can perform a range of tasks including Q&A, creative writing and programming support. PFN expects PLaMo Prime’s communicative abilities and wide knowledge base are suited for use in a variety of domains such as education, customer support and content production.

Pricing: Free (during the trial period)

Sign-ups for PLaMo API and PLaMo Chat are accepted on https://plamo.preferredai.jp/ (website in Japanese).

Going forward, PFN group will continue making improvements to PLaMo and offer solutions that cater to on-premise environments and specific business needs, thereby supporting companies across industries to implement innovative AI solutions for their competitive advantages.

About PLaMo™ Large Language Model

PLaMo (“plah-mo”) is a large language model built by the Preferred Networks group fully from scratch in Japan. Trained with the group’s proprietary high-quality dataset that include abundant Japanese content, PLaMo has achieved a world-class performance in Japanese-language tasks including financial benchmarks and the national medical examination. The PLaMo line of models include PLaMo Prime, the commercial flagship model, PLaMo Lite, the small language model for use in edge devices such as vehicles and production facilities, and PLaMo-100B – Pretrained (currently available with a non-commercial and commercial licenses), the model developed by PFE with support from GENIAC (Generative AI Accelerator Challenge), a project initiated by Japan’s Ministry of Economy, Trade and Industry and New Energy and Industrial Technology Development Organization (NEDO) for enhancing development of generative AI in Japan.