KMMLU-Pro
Solar Pro 4 79.2% leads since Aug 2026 In use Models answer multiple-choice questions taken from Korean National Professional Licensure exams. It tests professional knowledge used in Korean industry.
Released in the same paper as KMMLU-Redux. The Hugging Face dataset is gated.
In the authors’ words
The development of Large Language Models (LLMs) requires robust benchmarks that encompass not only academic domains but also industrial fields to effectively evaluate their applicability in real-world scenarios. In this paper, we introduce two Korean expert-level benchmarks. KMMLU-Redux, reconstructed from the existing KMMLU, consists of questions from the Korean National Technical Qualification exams, with critical errors removed to enhance reliability. KMMLU-Pro is based on Korean National Professional Licensure exams to reflect professional knowledge in Korea. Our experiments demonstrate that these benchmarks comprehensively represent industrial knowledge in Korea. We release our dataset publicly available.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Multiple choice
Solar Pro 4 leads with 79.2% among models released by Aug 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Aug 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 Solar Pro 4 79.2%
- 2 Solar Open 2 78.4%
- 3
K-EXAONE 2.0 69.1% - 4 EXAONE 4.0 32B 67.7%
- 5 EXAONE 4.5 33B 67.6%
- 6 K-EXAONE 67.3%
- 7 Solar Open 100B 64%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Solar Pro 4 | 79.2% | Aug 2026 | Model card | |
| 2 | Solar Open 2 | 78.4% | Jul 2026 | Model card | |
| 3 | K-EXAONE 2.0 | 69.1% | Jul 2026 | Model card | |
| 4 | EXAONE 4.0 32B | 67.7% | Jul 2025 | Model card | |
| 5 | EXAONE 4.5 33B | 67.6% | Apr 2026 | Model card | |
| 6 | K-EXAONE | 67.3% | Dec 2025 | Model card | |
| 7 | Solar Open 100B | 64% | Dec 2025 | Model card |
3 models have held number one since its release in Jul 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
2 labs put it in a launch post in the last six months
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 11 Aug 2026 | Solar Pro 4 | 79.2 | KMMLU-Pro* (in-house) |
| 31 Jul 2026 | K-EXAONE 2.0 | 69.1 | KMMLU-Pro |
| 22 Jul 2026 | Solar Open 2 | 78.4 | KMMLU-Pro |
| 9 Apr 2026 | EXAONE 4.5 33B | 67.6 | KMMLU-Pro |
| 31 Dec 2025 | Solar Open 100B | 64 | KMMLU-Pro |
| 31 Dec 2025 | K-EXAONE | 67.3 | KMMLU-Pro |
| 15 Jul 2025 | EXAONE 4.0 32B | 67.7 | KMMLU-Pro (reasoning) |
6 system cards report it, from 2 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 5 Aug 2026 | K-EXAONE 2.0 technical report | 69.1 | KMMLU-Pro (Table 6, Reasoning mode) | Capabilities |
| 22 Jul 2026 | Solar Open 2 technical report | 78.4 | KMMLU-Pro (Table 4) | Capabilities |
| 9 Apr 2026 | EXAONE 4.5 33B technical report | 67.6 | KMMLU-Pro (Table 3, Reasoning mode) | Capabilities |
| 11 Jan 2026 | Solar Open 100B technical report | 64 | KMMLU-Pro (Table 5) | Capabilities |
| 5 Jan 2026 | K-EXAONE technical report | 67.3 | KMMLU-Pro (Table 3, Reasoning mode). Non-reasoning 63.5 (Table 4) | Capabilities |
| 15 Jul 2025 | EXAONE 4.0 32B technical report | 67.7 | KMMLU-Pro (KO) (Table 3, Reasoning). Non-reasoning 60.0 (Table 4) | Capabilities |
More in Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects.
Intern-S2-397B 89.8% - SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy.
Qwen 3.6 Max (Preview) 73.9% - WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1.
Tencent Hy4 preview 83.9% - BIG-Bench Extra Hard In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is a harmonic mean of task accuracies.
Gemma 4 31B IT 74.4% - HLE-Verified In use Verified and revised version of Humanity's Last Exam (1,811 items) with uncertain items removed.
Seed2.0 Pro 73.6% - HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. GPT-6 Astra 65.8%