Upstage
We track 9 benchmarks that Upstage made or helped make. Its models are on its lab page.
1 in use 0 with public code 4 with a leaderboard here
Benchmarks
| Name | Kind | Released | Best score | Status |
|---|---|---|---|---|
| KBank-MMLU (Upstage) Models answer Korean finance questions in a multiple-choice format. Upstage built the set in house. | Finance | Jul 2025 | | Fading |
| Ko-AIME 2025 (Upstage) Models solve AIME 2025 competition problems in Korean. Upstage built the set in house. | Competition problems with answers | Jan 2026 | | Fading |
| Ko-GDPval Agents complete 170 Korean office tasks from 58 occupations in 11 industries and produce deliverable files. Rubrics grade data, content, structure, quality and format. | Tasks from real jobs | Jul 2026 | | In use |
| Ko-Arena-Hard v2 (Upstage) Models answer hard user prompts in Korean, judged against a baseline in the style of Arena-Hard v2. Upstage built it in house. | Chat and instructions in many languages | Dec 2025 | | Fading |
| Ko-IFEval (Upstage) Models follow verifiable instructions written in Korean. Upstage adapted IFEval to Korean for the Open Ko-LLM Leaderboard2. | Chat and instructions in many languages | Oct 2024 | 64% | System cards only |
| Ko-AIME 2024 (Upstage) Models solve AIME 2024 competition problems translated into Korean. Upstage built the set in house. | Competition problems with answers | Dec 2025 | | System cards only |
| Open Ko-LLM Leaderboard 2 Models complete nine Korean evaluations covering reasoning, instructions, social values, safety, and other abilities. The organizers score submitted models on private test data. | Reasoning in many languages | Aug 2024 | 49.6% | Not in cards |
| Open Ko-LLM Leaderboard Models answer Korean questions about knowledge, truthfulness, and commonsense reasoning. The leaderboard uses translated tasks alongside a Korean commonsense benchmark. | Knowledge in many languages | Sep 2023 | | Not in cards |
| DP-Bench Document parsers and vision models convert PDFs and scanned pages into structured HTML or Markdown. Output is compared with reference structure. | Documents, charts and text in images | Oct 2024 | Not in cards |