KorNAT
Models answer questions about Korean social values and common knowledge. The authors compare answers with Korean reference judgments.
The Open Ko-LLM Leaderboard reports knowledge and social value scores separately.
- Made by
- Jiyoung Lee and colleagues
- Released
- February 2024
- Kind
- Knowledge in many languages
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Language
- Korean
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- A host
- Strongest model tested
- Claude Opus 5 #11 on the capability index
- Last model added
- August 2026
- Scores refreshed
- November 20th, 2026
Claude Opus 5 leads with 65.4% among models released by August 2026 Claude Opus 5 leads with 65.4% among models released by August 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The leaderboard reports no overall score, so this is the plain mean of Social Values and Common Knowledge. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in August 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 14 more
- 65.4 Claude Opus 5
- 64.8 GPT-5.6 Sol
- 63.4
Solar Pro 4
- 63.2
Qwen 3.8 Max
- 62.5 Kimi K3
- 62.0 Grok 4.6
- 61.9 GPT-5.6 Luna
- 60.5 naver-hyperclovax/HyperCLOVAX-SEED-Think-32B
- 60.0
Gemini 3.1 Pro
- 59.0
EXAONE 4.0 32B
- 58.4
Gemma 3 27B
- 58.2 GLM-5.2
- 57.5
DeepSeek-V4-Pro
- 57.4
Qwen3-32B
- 57.4
Gemini 3.6 Flash
- 57.0
EXAONE 4.5 33B
- 56.4
Llama 3.3 70B Instruct
- 56.1 Claude Haiku 4.5
- 50.8 skt/A.X-4.0-Light
- 48.6 kakaocorp/kanana-1.5-8b-instruct-2505
- 1 Claude Opus 5 ✓ 65.4%
- 2 GPT-5.6 Sol ✓ 64.8%
- 3
Solar Pro 4 ✓ 63.4%
- 4
Qwen 3.8 Max ✓ 63.2%
- 5 Kimi K3 ✓ 62.5%
- 6 Grok 4.6 ✓ 62%
- 7 GPT-5.6 Luna ✓ 61.9%
- 8 naver-hyperclovax/HyperCLOVAX-SEED-Think-32B ✓ 60.5%
- 9
Gemini 3.1 Pro ✓ 60%
- 10
EXAONE 4.0 32B ✓ 59%
- 11
Gemma 3 27B ✓ 58.4%
- 12 GLM-5.2 ✓ 58.2%
- 13
DeepSeek-V4-Pro ✓ 57.5%
- 14
Qwen3-32B ✓ 57.4%
- 15
Gemini 3.6 Flash ✓ 57.4%
- 16
EXAONE 4.5 33B ✓ 57%
- 17
Llama 3.3 70B Instruct ✓ 56.4%
- 18 Claude Haiku 4.5 ✓ 56.1%
- 19 skt/A.X-4.0-Light ✓ 50.8%
- 20 kakaocorp/kanana-1.5-8b-instruct-2505 ✓ 48.6%
Source: Official KorNAT leaderboard
- 65.4 Claude Opus 5
- 64.8 GPT-5.6 Sol
- 63.4
Solar Pro 4
- 63.2
Qwen 3.8 Max
- 62.5 Kimi K3
- 62.0 Grok 4.6
- 61.9 GPT-5.6 Luna
- 60.0
Gemini 3.1 Pro
- 59.0
EXAONE 4.0 32B
- 58.4
Gemma 3 27B
- 58.2 GLM-5.2
- 57.5
DeepSeek-V4-Pro
- 57.4
Qwen3-32B
- 57.4
Gemini 3.6 Flash
- 57.0
EXAONE 4.5 33B
- 56.4
Llama 3.3 70B Instruct
- 56.1 Claude Haiku 4.5
- 1 Claude Opus 5 ✓ 65.4%
- 2 GPT-5.6 Sol ✓ 64.8%
- 3
Solar Pro 4 ✓ 63.4%
- 4
Qwen 3.8 Max ✓ 63.2%
- 5 Kimi K3 ✓ 62.5%
- 6 Grok 4.6 ✓ 62%
- 7 GPT-5.6 Luna ✓ 61.9%
- 8
Gemini 3.1 Pro ✓ 60%
- 9
EXAONE 4.0 32B ✓ 59%
- 10
Gemma 3 27B ✓ 58.4%
- 11 GLM-5.2 ✓ 58.2%
- 12
DeepSeek-V4-Pro ✓ 57.5%
- 13
Qwen3-32B ✓ 57.4%
- 14
Gemini 3.6 Flash ✓ 57.4%
- 15
EXAONE 4.5 33B ✓ 57%
- 16
Llama 3.3 70B Instruct ✓ 56.4%
- 17 Claude Haiku 4.5 ✓ 56.1%
Source: Official KorNAT leaderboard
| # | Entry | Score | Setting | Submitted by | Released | Run on | Source |
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5 ✓ | 65.4% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 2 | GPT-5.6 Sol ✓ | 64.8% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 3 | | 63.4% | organizers | Aug 2026 | Aug 2026 | Official leaderboard | |
| 4 | | 63.2% | organizers | Aug 2026 | Aug 2026 | Official leaderboard | |
| 5 | Kimi K3 ✓ | 62.5% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 6 | Grok 4.6 ✓ | 62% | organizers | Aug 2026 | Aug 2026 | Official leaderboard | |
| 7 | GPT-5.6 Luna ✓ | 61.9% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 8 | naver-hyperclovax/HyperCLOVAX-SEED-Think-32B ✓ | 60.5% | organizers | Aug 2026 | Official leaderboard | ||
| 9 | | 60% | organizers | Feb 2026 | Aug 2026 | Official leaderboard | |
| 10 | | 59% | organizers | Jul 2025 | Aug 2026 | Official leaderboard | |
| 11 | | 58.4% | organizers | Mar 2025 | Aug 2026 | Official leaderboard | |
| 12 | GLM-5.2 ✓ | 58.2% | organizers | Jun 2026 | Aug 2026 | Official leaderboard | |
| 13 | | 57.5% | organizers | Apr 2026 | Aug 2026 | Official leaderboard | |
| 14 | | 57.4% | organizers | Apr 2025 | Aug 2026 | Official leaderboard | |
| 15 | | 57.4% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 16 | | 57% | organizers | Apr 2026 | Aug 2026 | Official leaderboard | |
| 17 | | 56.4% | organizers | Dec 2024 | Aug 2026 | Official leaderboard | |
| 18 | Claude Haiku 4.5 ✓ | 56.1% | organizers | Oct 2025 | Aug 2026 | Official leaderboard | |
| 19 | skt/A.X-4.0-Light ✓ | 50.8% | organizers | Aug 2026 | Official leaderboard | ||
| 20 | kakaocorp/kanana-1.5-8b-instruct-2505 ✓ | 48.6% | organizers | Aug 2026 | Official leaderboard |
| # | Model | Score | Setting | Submitted by | Released | Run on | Source |
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5 ✓ | 65.4% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 2 | GPT-5.6 Sol ✓ | 64.8% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 3 | | 63.4% | organizers | Aug 2026 | Aug 2026 | Official leaderboard | |
| 4 | | 63.2% | organizers | Aug 2026 | Aug 2026 | Official leaderboard | |
| 5 | Kimi K3 ✓ | 62.5% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 6 | Grok 4.6 ✓ | 62% | organizers | Aug 2026 | Aug 2026 | Official leaderboard | |
| 7 | GPT-5.6 Luna ✓ | 61.9% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 8 | | 60% | organizers | Feb 2026 | Aug 2026 | Official leaderboard | |
| 9 | | 59% | organizers | Jul 2025 | Aug 2026 | Official leaderboard | |
| 10 | | 58.4% | organizers | Mar 2025 | Aug 2026 | Official leaderboard | |
| 11 | GLM-5.2 ✓ | 58.2% | organizers | Jun 2026 | Aug 2026 | Official leaderboard | |
| 12 | | 57.5% | organizers | Apr 2026 | Aug 2026 | Official leaderboard | |
| 13 | | 57.4% | organizers | Apr 2025 | Aug 2026 | Official leaderboard | |
| 14 | | 57.4% | organizers | Jul 2026 | Aug 2026 | Official leaderboard | |
| 15 | | 57% | organizers | Apr 2026 | Aug 2026 | Official leaderboard | |
| 16 | | 56.4% | organizers | Dec 2024 | Aug 2026 | Official leaderboard | |
| 17 | Claude Haiku 4.5 ✓ | 56.1% | organizers | Oct 2025 | Aug 2026 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Social Values, Common Knowledge and Common Sense. Gemini 3.1 Pro leads 8 of them, and EXAONE 4.0 32B leads the other one. Arrows show how far a model moved from its overall rank among the models we track.
Social Values
17 models we track have a score here.
- 1
EXAONE 4.0 32B ↑8 33.3%
- 2 Claude Opus 5 ↓1 33%
- 3 GPT-5.6 Sol ↓1 32%
- 4
Qwen 3.8 Max 30.6%
- 5
Solar Pro 4 ↓2 30.4%
- 6
Gemma 3 27B ↑4 30.3%
- 7 Kimi K3 ↓2 29.9%
- 8 Grok 4.6 ↓2 29.2%
- 9 GPT-5.6 Luna ↓2 29.1%
- 10
Qwen3-32B ↑3 29%
Common Knowledge
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 98.5%
- 2
Gemini 3.6 Flash ↑12 98%
- 3 Claude Opus 5 ↓2 97.8%
- 4 GPT-5.6 Sol ↓2 97.6%
- 5
Solar Pro 4 ↓2 96.4%
- 6
Qwen 3.8 Max ↓2 95.8%
- 7 Kimi K3 ↓2 95.2%
- 8 Grok 4.6 ↓2 94.9%
- 9 GPT-5.6 Luna ↓2 94.7%
- 10 GLM-5.2 ↑1 91.1%
Common Sense
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 97.9%
- 2 GPT-5.6 Sol 97.1%
- 3
Gemini 3.6 Flash ↑11 97%
- 4 Claude Opus 5 ↓3 96.9%
- 5
Solar Pro 4 ↓2 96.2%
- 6
Qwen 3.8 Max ↓2 95.9%
- 7 Kimi K3 ↓2 95.6%
- 8 GPT-5.6 Luna ↓1 95.4%
- 9 Grok 4.6 ↓3 92.7%
- 10
DeepSeek-V4-Pro ↑2 92.6%
English
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 98.9%
- 2 Claude Opus 5 ↓1 98.4%
- 3
Gemini 3.6 Flash ↑11 98.2%
- 4 GPT-5.6 Sol ↓2 97.8%
- 5
Qwen 3.8 Max ↓1 97.7%
- 6 Kimi K3 ↓1 97.4%
- 7
Solar Pro 4 ↓4 97.2%
- 8 GLM-5.2 ↑3 96.9%
- 9 Grok 4.6 ↓3 96.2%
- 10
DeepSeek-V4-Pro ↑2 95.2%
Korean
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 98.7%
- 2
Gemini 3.6 Flash ↑12 98.1%
- 3 Claude Opus 5 ↓2 97.9%
- 4 GPT-5.6 Sol ↓2 97.6%
- 5 GPT-5.6 Luna ↑2 95.4%
- 6
Qwen 3.8 Max ↓2 95.3%
- 7
Solar Pro 4 ↓4 95.3%
- 8 Kimi K3 ↓3 94.1%
- 9 Grok 4.6 ↓3 92.8%
- 10
EXAONE 4.5 33B ↑5 92.2%
Korean History
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 97.4%
- 2
Gemini 3.6 Flash ↑12 96.3%
- 3 GPT-5.6 Sol ↓1 96.2%
- 4 Claude Opus 5 ↓3 95.9%
- 5 GPT-5.6 Luna ↑2 95.1%
- 6
Solar Pro 4 ↓3 92.7%
- 7 Grok 4.6 ↓1 91.3%
- 8 Claude Haiku 4.5 ↑9 88%
- 9
EXAONE 4.5 33B ↑6 87.9%
- 10
Qwen 3.8 Max ↓6 87.7%
Mathematics
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 98.9%
- 2
Gemini 3.6 Flash ↑12 98.8%
- 3 Kimi K3 ↑2 98.6%
- 4 Claude Opus 5 ↓3 98.6%
- 5
Qwen 3.8 Max ↓1 98.5%
- 6 Grok 4.6 98%
- 7 Claude Haiku 4.5 ↑10 97.9%
- 8
DeepSeek-V4-Pro ↑4 97.9%
- 9
Solar Pro 4 ↓6 97.9%
- 10 GLM-5.2 ↑1 97.8%
Science
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 99.2%
- 2 Claude Opus 5 ↓1 99.1%
- 3 GPT-5.6 Sol ↓1 98.8%
- 4
Gemini 3.6 Flash ↑10 98.8%
- 5
Qwen 3.8 Max ↓1 98.8%
- 6 GPT-5.6 Luna ↑1 98.5%
- 7
Solar Pro 4 ↓4 98.5%
- 8 Grok 4.6 ↓2 98.2%
- 9 Kimi K3 ↓4 98%
- 10
DeepSeek-V4-Pro ↑2 97.2%
Social Studies
17 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 98.8%
- 2
Gemini 3.6 Flash ↑12 98.8%
- 3 GPT-5.6 Sol ↓1 98.2%
- 4
Qwen 3.8 Max 98.2%
- 5 Claude Opus 5 ↓4 98%
- 6 GPT-5.6 Luna ↑1 97.4%
- 7 Kimi K3 ↓2 97.4%
- 8
Solar Pro 4 ↓5 97.2%
- 9 Grok 4.6 ↓3 96%
- 10
Llama 3.3 70B Instruct ↑6 94.5%
6 models have held number one on the official leaderboard since its release in February 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official KorNAT leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official KorNAT leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Multilingual
- MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
- WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET.
K-EXAONE 90.5%
- MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language.
Qwen3.7-Max 87%
- CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories.
Solar Open 2 90.7%
- Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
- MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%
Data audit
All 10 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the official leaderboard (20 entries).
- Pass Models scored 17 models we track, plus 3 other entries on its leaderboard.
- Pass Newest score The newest is from August 2026, for Grok 4.6.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 5 (#11).
- Pass Headroom The best is 65.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 17 rows for models we track and 3 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.