Hero Cover
Newsroom
GoTo Releases New 27 Billion Parameter Open-Source LLM Trained for Indonesia
19 October 2026
title

The 27 billion parameter model, announced at Bali Tech Week, is open for universities, startups, businesses and public institutions to use and adapt

 
SANUR, Bali, 19 October 2026 – PT GoTo Gojek Tokopedia Tbk (“GoTo” or "the Company", IDX: GOTO) today released a 27 billion parameter large language model (LLM) trained for Bahasa Indonesia as well as the regional languages Javanese, Sundanese, Balinese and Bataknese.
 
The model, which was launched by the Company at Bali Tech Week, was created by GoTo’s extensive team of AI engineers and builds on open-model work GoTo began in 2024. It is now freely available for startups, businesses, universities, researchers and public institutions to use, study and adapt, allowing them to build locally relevant AI without starting from zero.
 
Large language models are the technology behind AI tools that answer questions, summarize documents and generate translations. They learn how language works from large, diverse amounts of text, and the more relevant and extensive the data they learn from, the better the results.
"Indonesia makes great use of global AI, but we must build our own too, to solve the problems that are unique to us," said Hans Patuwo, GoTo Group CEO. "AI created for global use may not be fully optimized for our localized data or the regional languages so many Indonesians grow up using at home. At GoTo, we are overcoming that challenge by building locally relevant AI for use with our own customers. Opening this technology to everyone is our way of giving back to the country, providing anyone building AI in Indonesia with a starting point rooted in the local context."
 
The open-sourcing of GoTo’s 27 billion parameter LLM supports the Ministry of Communication and Digital Affairs’ (Komdigi) Indonesia Digital Vision 2045, which aims to create an inclusive, collaborative and innovation-driven digital ecosystem for a sovereign, advanced and sustainable Indonesia – and to move the country from consumer to creator of AI technology.
 

Built from Indonesian data

The model was trained on Indonesian-language text from media partners and public-sector websites, alongside regional-language data. For Bataknese, GoTo worked with Universitas Sumatera Utara (USU) to collect data that reflects how the language is used today.

At 27 billion parameters, it is a mid-sized model. That makes it capable enough for demanding language tasks, yet small enough for a university, startup or public institution to run on its own servers rather than in a large data center.

Developing Indonesia's AI builders

GoTo is also investing in the people who will build the next generation of Indonesian AI. Interns from seven universities, from Sumatera to Sulawesi and including Udayana University in Bali, have worked alongside GoTo’s AI teams on real projects, from preparing data and training models to testing them. They also lent their voices to train GoTo’s Indonesian speech technology and helped gather Batak-language data.

Together, GoTo’s efforts reflect a broader approach to strengthening Indonesia’s AI capabilities: building technology around local needs, opening it for others to build on, and helping more Indonesians develop the skills to build AI themselves.

GoTo's 27 billion parameter model can be accessed at:

https://huggingface.co/GoToCompany/gemma3-27b-cpt-gotoai-v3-instruct

About GoTo Group

GoTo is the largest digital ecosystem in Indonesia. GoTo's mission is to 'empower progress' by offering technology infrastructure and solutions that help everyone to access and thrive in the digital economy.
The GoTo ecosystem provides a wide range of services, including mobility, delivery, payments, financial services, and technology solutions for merchants. The ecosystem also provides e-commerce services through Tokopedia and banking services through its partnership with Bank Jago.

 
Powered By
goto