GoTo Group and Indosat Ooredoo Hutchison introduced the upgraded Sahabat-AI 70 billion parameter model, along with a new multilingual AI chat service.

This marks a major milestone in Indonesia’s open-source LLM development, reinforcing digital sovereignty and national innovation.

Initially launched during Indosat’s Indonesia AI Day in November 2024, Sahabat-AI now powers a more capable and accurate chat interface accessible via sahabat-ai.com and under “Popular Services” in the GoPay app.

“With the launch of our 70 billion parameter model and the new chat service, Sahabat-AI takes a major leap forward in building a uniquely Indonesian AI ecosystem,” said Patrick Walujo, CEO of GoTo Group.

Walujo emphasized that Sahabat-AI supports President Prabowo’s vision for domestically developed and hosted technology while providing smarter, faster, and more cost-effective AI services to users across the country.

A National Model Built on Local Languages and Infrastructure

The Sahabat-AI model now supports five local languages Bahasa Indonesia, Javanese, Sundanese, Balinese, and Bataknese alongside various international languages.

Built on Indosat’s GPU Merdeka infrastructure, the model is optimized to run efficiently on Indonesian-based cloud systems, ensuring both regulatory compliance and broad accessibility.

“Indosat is proud to lead the development of Indonesia’s sovereign AI capabilities,” said Vikram Sinha, President Director and CEO of Indosat Ooredoo Hutchison.

He described Sahabat-AI not just as a technical asset but a “national asset powered by collaboration”, supported by real-time multilingual performance and data residency policies, making it viable for public sector use and enterprise-level deployment.

Strategic Collaborations Strengthen Indonesia’s AI Ecosystem

Sahabat-AI’s development reflects the Indonesian philosophy of gotong royong (mutual cooperation).

It is the result of joint efforts among universities, media houses, and government institutions, including partnerships with the University of Indonesia, Gadjah Mada University, Bandung Institute of Technology, and media organizations such as Kompas Group and Tempo.

Luhut Binsar Pandjaitan, Chairman of the National Economic Council, said: “Data sovereignty isn’t just a technical matter, it’s a matter of national independence in the digital era.”

He praised GoTo and Indosat for embedding cultural and linguistic relevance into Sahabat-AI, aligning it with Indonesia’s digital transformation agenda.

Empowering Future AI Engineers Through Internships

A structured internship program plays a key role in Sahabat-AI’s development strategy.

University students contribute directly to refining the model, gaining rare practical experience in dataset preprocessing, architecture design, and model training.

Komang Ayu, an intern from the University of Udayana, said: “This experience deepened my understanding of the end-to-end development of large language models.”

The internship program is part of Sahabat-AI’s broader goal to grow local AI expertise, making the platform both an innovation hub and a training ground for Indonesia’s next generation of engineers.

Open Access for Developers and Public Use

Since the debut of its earlier 8B and 9B models, Sahabat-AI has been downloaded over 35,000 times on Hugging Face.

The latest 70B model is available through the official Sahabat-AI website and its Hugging Face page, supporting a diverse range of users from startups and researchers to government bodies.

By keeping the model open-source and locally hosted, Sahabat-AI encourages experimentation, scalability, and innovation all within Indonesia’s legal and cultural boundaries.

PHOTO: IOH

This article was created with AI assistance.