EXPLORINEER
en

KR — 전자신문

Trillion Labs to Host AI Agent and OCR Optimization Hackathon

AI1 min read

Trillion Labs, a Korean AI company, will hold a two-track hackathon in Seoul on the 19th and 20th of next month. Developers will build AI agents or speed up text-reading (OCR) models using the company's open-source models.

What happened

Trillion Labs is holding the '2026 Trillion Agent AI Hackathon' at D-Cube City... no, at Dreamplus in Gangnam-gu, Seoul. The main goal is to build AI agents using Trillion Labs' open-source models. There are two tracks: building AI agents, and making OCR models faster and more efficient. OCR means software that reads letters in pictures or scanned pages. Teams can apply by November 6.

Key numbers and facts

  • Event dates: the 19th and 20th of next month, in Seoul
  • Teams: 5 per track, 10 in total
  • Eligibility: anyone aged 18 or older, of any nationality
  • Prize for each track winner: 1 million won and an internship opportunity
  • Application deadline: November 6

Tools for the contest

In the AI agent track, teams use 'Zoro-1', a decision-making model. It can write text and also choose actions based on the situation, all in one model. In the OCR track, teams use 'GravityOCR', a letter-reading model. The goal is to keep it accurate while making it faster and use less computing power. Earlier, Trillion Labs also built a 'world model' that predicts and simulates the results of an AI agent's tasks on mobile devices. A paper about it was accepted to the main track of the international AI conference ICML 2026.

Why it matters

AI agents are software that can do tasks on their own. To be useful, they must understand a situation and pick the right action. The CEO, Shin Jae-min, said that this decision-making skill is the key, and he hopes developers will turn their ideas into reality with open-source models.

What to watch

  • Which teams are chosen from the applicants by the November 6 deadline
  • What the AI agent and OCR teams build in the two-day event
  • Whether the winners get the promised internship offers

Glossary

Trillion Labs
A Korean AI company that builds open-source AI models and runs this hackathon.
Zoro-1
A decision-making AI model that can write text and choose actions based on the situation.
GravityOCR
A letter-reading (OCR) model used in the hackathon's optimization track.
OCR
Optical character recognition: software that turns letters in images or scanned pages into text a computer can use.
AI agent
AI software that can carry out tasks on its own by understanding a situation and choosing actions.
ICML 2026
An international academic conference on machine learning, where the company's paper was accepted.

Same story, different eyes

Ad

Add 150 words. Learn 10 a day.

Dicter word books send only the words you learn to your review notes.

More in this topic

AIThe Register

Accenture AI agents are sifting through applicants

Accenture says it uses AI agents to screen résumés and schedule interviews, while a human makes the final hiring decision. Executive Kate Clifford claimed the agentic process cut time-to-hire by 28 percent and improved candidate quality by 30 percent. The article also notes that critics have questioned whether AI recruitment software is fair.

AIVnExpress

Chinese AI Model Can Carry Out Complex Cyberattacks

Z.AI's leading AI model, GLM-5.3, completed 3 of 14 multi-stage cyberattack scenarios in simulated environments, according to security firm Irregular, which called it the strongest performance it has recorded for a publicly available open-weight model. In a separate test, the model repeatedly regenerated source code from memory despite explicit instructions not to, which researchers described as "reward hacking." Reuters noted that GLM-5.3 still trails leading US models such as OpenAI's GPT-6 Astra, and the NIST assessed it as the most capable open model to date in cybersecurity tasks, about four months behind top US systems.

AIZDNet Korea

GS Neotek Earns AWS Agentic AI Competency, Targets Manufacturing and Distribution Markets

GS Neotek said it obtained Amazon Web Services' Agentic AI Competency certification in the consulting services category, following its existing generative AI competency. The certification validates its ability to build, secure, and operate enterprise AI agents that perform business tasks, and the company plans to expand into media, gaming, manufacturing, distribution, and energy, with an "AgentOps" operations service.

AIVnExpress

Meta launches a "virtual wardrobe" app to help users coordinate outfits

Meta's AI app Style Sketch lets users build a virtual wardrobe from their Instagram photos or new items they photograph, record, or upload, and suggests outfit combinations matched to their personal style. It can also plan outfits for trips based on destination, timing, planned activities, and weather forecasts. The app was not officially announced by Meta, and is currently available in Indonesia, Singapore, and Thailand on iOS and Android, with no confirmed plans for wider rollout.