Allganize backs Upstage-led bid to take Korea's independent AI foundation model global
Allganize said the consortium led by Upstage passed the first round of evaluations for the government's "independent AI foundation model project," and it will support the group in scaling the model internationally. The initiative, run by the Ministry of Science and ICT, aims to build a domestically developed base model and a globally competitive AI ecosystem. Three consortia advanced: Upstage, LG AI Research Institute, and SK Telecom. Allganize has joined the Upstage consortium.
What the consortium is building
The team is developing "Solar WBL," a foundation model intended to spread across core industries: finance, health care, manufacturing, law, the public sector, and education. The goal is clear: push AI into real operations across sectors while securing technological sovereignty.
Allganize's role
Allganize will lead country- and industry-specific optimization and drive global market rollout. The company plans to apply its agent retrieval-augmented generation (RAG) stack and LLM fine-tuning experience so the models work within each region's language, data rules, and industry workflows. The focus is practical adoption, not just benchmarks.
Allganize has shipped AI in high-security environments across finance and the public sector. In Korea, it provides its all-in-one LLM solution "Alli" to Woori Investment & Securities, KB Securities, Korea Land & Housing Corporation (LH), and KEPCO KDN (KDN). It also serves Nomura Securities in Japan and the State Government of Oklahoma in the United States.
"Based on Allganize's accumulated experience supplying AI solutions and its technical capabilities, we will do our best to ensure the AI foundation model developed through this project does not stop at technology development but gains real competitiveness in the global market," said Lee Won-gang, deputy CEO of Allganize.
What this means for IT and development teams
- Expect an industry-focused foundation model (Solar WBL) with pathways to deploy in finance, health care, manufacturing, law, public, and education scenarios.
- Country-level optimization implies attention to local regulations, data residency, and multilingual performance-plan for evaluation and red-teaming by region.
- RAG + fine-tuning will be central: invest in high-quality knowledge bases, retrieval pipelines, and continuous evaluation to reduce hallucinations and latency.
- Past deployments in security-sensitive orgs suggest integration patterns that account for access control, auditability, and model governance from day one.
Learn more
Project context: Ministry of Science and ICT | Consortium lead: Upstage
Upskilling for teams working with RAG, evals, and LLM fine-tuning: Complete AI Training - courses by job
※ This article has been translated by AI. View Original Article (Korean)
Your membership also unlocks:
