#AI Optimization #Metareal GP #Rosetta Inc.

Introduction of Metareal Green Prompt (Metareal GP)

On October 3, 2025, Rosetta Inc., under the leadership of Junichi Goishi, will commence offering the premium plan for the Metareal Green Prompt (Metareal GP), an AI agent designed to optimize GPU power costs and CO₂ emissions arising from the operations of large language models (LLMs).

This subscription-based service will be available for a monthly fee of 50,000 yen (tax included), allowing users to utilize the service up to 20 times a day or 200 times a month. The Metareal GP leverages cutting-edge AI technology to balance energy consumption within user-accepted waiting times, ensuring both effective output and sustainable operation.

Development Background

In the realm of SaaS-based LLM services, the struggle to balance environmental sustainability with service quality (Service Level Agreement, SLA) continues to be a significant challenge. While there is a strong demand for stabilizing energy loads and promoting nighttime operations, the risk of customer churn due to processing delays remains a crucial factor. Traditional scheduling systems typically prioritize power costs alone, making it challenging to implement flexible controls that consider CO₂ intensity and future carbon pricing. Consequently, there has been a growing need for tools that balance LLM response quality while maximizing environmental and economic value.

Key Features of Metareal GP

The Metareal GP boasts four primary features that set it apart:

1. Maximized Power Allocation for Each Prompt: Utilizing regional GPU cluster usage rates and power CO₂ intensity, along with power and carbon taxes, the AI dynamically optimizes load distribution.

2. Sustainable Operations without Compromising SLA: It visualizes and minimizes environmental costs without negatively impacting the service quality.

3. Rapid 5-Second Redistributions: With industry's best-in-class granularity, the service can respond immediately to changes in processing demands and generate CSV/PPT outputs for internal reporting automatically.

4. Differentiation from Existing Schedulers: It incorporates not just power pricing but also CO₂ intensity and Red-Amber-Green (RAG) taxation indicators into its decision-making processes.

This service is designed for:

• - CEOs of LLM SaaS Companies: To differentiate products with environmental considerations and enhance investment decision-making.
• - Operations VPs: To strengthen field decision-making that pursues environmental optimization and cost reduction in infrastructure operations.

Strengthening Industry-Specific AI Consulting with Metareal AI's Generative AI Series

Rosetta is committed to rolling out a series of generative AI solutions tailored to various industries, beginning with Metareal DD targeted at the financial sector. Each deployment aims to address specific challenges faced in different sectors. By combining our AI consulting expertise, we can facilitate faster and more reliable AI implementations.

Our competitive edge in consulting lies in our ability to deliver 'AI that is usable and desirable in the field' at an unmatched speed. Despite the rapid evolution of AI technology, the critical factor remains how companies can harness these innovations to create value in practice. However, operational teams typically prioritize maintaining the status quo, leading them to adopt new technologies cautiously. Therefore, our consulting model emphasizes 'how to rapidly iterate and adapt to the field' rather than merely presenting the 'correct solution.' We've successfully implemented AI solutions at an unprecedented pace in industries, including media, where adoption was previously considered challenging.

Features of Metareal AI Consulting

1. Specialization Over Generalization: With experience working with over 6,000 companies across 2,000 distinct fields, we offer industry-leading specialized knowledge able to develop AI solutions immediately applicable to specific operational tasks.
2. Collaboration of Multiple LLMs: By utilizing our proprietary