LLM Benchmark: Danske Idiomer
Models match Danish idioms to pictures, complete expressions and write definitions. GPT-4.1 grades definitions, and answer keys cover the other tasks.
The maker compares 20 models across seven question types. Eleven models answer all 200 questions. Nine text-only models skip the 30 pictures.
- Made by
- idiom.dk
- Released
- January 2026
- Kind
- Reading in many languages
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Language
- Danish
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision
- Strongest model tested
- GPT-5.2 #51 on the capability index
- Last model added
- December 2025
- Scores refreshed
- November 20th, 2026
GPT-5.1 leads with 97.5% among models released by December 2025 GPT-5.1 leads with 97.5% among models released by December 2025
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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 December 2025, 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 16 more
- 97.5 GPT-5.1
- 96.5 GPT-5.2
- 93.0 GPT-4o mini
- 93.0 Grok 4
- 90.0 Claude Sonnet 4
- 87.0
Mistral Large 3
- 83.5 DeepSeek Reasoner
- 82.9
DeepSeek-V3.2
- 82.3 Mistral Large 2.1
- 81.2 GPT-3.5 Turbo
- 81.0 Claude 3 Haiku
- 80.5
Gemma 3 12B
- 80.5 Grok 4.1 Fast
- 80.0
Llama 3.3 70B
- 75.3
Gemma 2 9B
- 72.5 GPT-5 mini
- 67.0
Magistral Small 1.2
- 63.5
Qwen3-8B
- 50.0 davinci-002
- 39.4 GPT-5 nano
- 1 GPT-5.1 97.5%
- 2 GPT-5.2 96.5%
- 3 GPT-4o mini 93%
- 4 Grok 4 93%
- 5 Claude Sonnet 4 90%
- 6
Mistral Large 3 87%
- 7 DeepSeek Reasoner 83.5%
- 8
DeepSeek-V3.2 82.9%
- 9 Mistral Large 2.1 82.3%
- 10 GPT-3.5 Turbo Jan 2024 81.2%
- 11 Claude 3 Haiku 81%
- 12
Gemma 3 12B 80.5%
- 13 Grok 4.1 Fast 80.5%
- 14
Llama 3.3 70B 80%
- 15
Gemma 2 9B 75.3%
- 16 GPT-5 mini 72.5%
- 17
Magistral Small 1.2 67%
- 18
Qwen3-8B 63.5%
- 19 davinci-002 50%
- 20 GPT-5 nano 39.4%
Source: Official LLM Benchmark: Danske Idiomer leaderboard
- 97.5 GPT-5.1
- 96.5 GPT-5.2
- 93.0 GPT-4o mini
- 93.0 Grok 4
- 90.0 Claude Sonnet 4
- 87.0
Mistral Large 3
- 82.9
DeepSeek-V3.2
- 81.2 GPT-3.5 Turbo
- 81.0 Claude 3 Haiku
- 80.5
Gemma 3 12B
- 80.5 Grok 4.1 Fast
- 80.0
Llama 3.3 70B
- 75.3
Gemma 2 9B
- 72.5 GPT-5 mini
- 67.0
Magistral Small 1.2
- 63.5
Qwen3-8B
- 50.0 davinci-002
- 39.4 GPT-5 nano
- 1 GPT-5.1 97.5%
- 2 GPT-5.2 96.5%
- 3 GPT-4o mini 93%
- 4 Grok 4 93%
- 5 Claude Sonnet 4 90%
- 6
Mistral Large 3 87%
- 7
DeepSeek-V3.2 82.9%
- 8 GPT-3.5 Turbo Jan 2024 81.2%
- 9 Claude 3 Haiku 81%
- 10
Gemma 3 12B 80.5%
- 11 Grok 4.1 Fast 80.5%
- 12
Llama 3.3 70B 80%
- 13
Gemma 2 9B 75.3%
- 14 GPT-5 mini 72.5%
- 15
Magistral Small 1.2 67%
- 16
Qwen3-8B 63.5%
- 17 davinci-002 50%
- 18 GPT-5 nano 39.4%
Source: Official LLM Benchmark: Danske Idiomer leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.1 | 97.5% | Nov 2025 | Official leaderboard | |
| 2 | GPT-5.2 | 96.5% | Dec 2025 | Official leaderboard | |
| 3 | GPT-4o mini | 93% | Jul 2024 | Official leaderboard | |
| 4 | Grok 4 | 93% | Jul 2025 | Official leaderboard | |
| 5 | Claude Sonnet 4 | 90% | May 2025 | Official leaderboard | |
| 6 | | 87% | Dec 2025 | Official leaderboard | |
| 7 | DeepSeek Reasoner | 83.5% | Official leaderboard | ||
| 8 | | 82.9% | Dec 2025 | Official leaderboard | |
| 9 | Mistral Large 2.1 | 82.3% | Official leaderboard | ||
| 10 | GPT-3.5 Turbo Jan 2024 | 81.2% | Jan 2024 | Official leaderboard | |
| 11 | Claude 3 Haiku | 81% | Mar 2024 | Official leaderboard | |
| 12 | | 80.5% | Mar 2025 | Official leaderboard | |
| 13 | Grok 4.1 Fast | 80.5% | Nov 2025 | Official leaderboard | |
| 14 | | 80% | Dec 2024 | Official leaderboard | |
| 15 | | 75.3% | Jun 2024 | Official leaderboard | |
| 16 | GPT-5 mini | 72.5% | Aug 2025 | Official leaderboard | |
| 17 | | 67% | Sep 2025 | Official leaderboard | |
| 18 | | 63.5% | Apr 2025 | Official leaderboard | |
| 19 | davinci-002 | 50% | Mar 2022 | Official leaderboard | |
| 20 | GPT-5 nano | 39.4% | Aug 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.1 | 97.5% | Nov 2025 | Official leaderboard | |
| 2 | GPT-5.2 | 96.5% | Dec 2025 | Official leaderboard | |
| 3 | GPT-4o mini | 93% | Jul 2024 | Official leaderboard | |
| 4 | Grok 4 | 93% | Jul 2025 | Official leaderboard | |
| 5 | Claude Sonnet 4 | 90% | May 2025 | Official leaderboard | |
| 6 | | 87% | Dec 2025 | Official leaderboard | |
| 7 | | 82.9% | Dec 2025 | Official leaderboard | |
| 8 | GPT-3.5 Turbo Jan 2024 | 81.2% | Jan 2024 | Official leaderboard | |
| 9 | Claude 3 Haiku | 81% | Mar 2024 | Official leaderboard | |
| 10 | | 80.5% | Mar 2025 | Official leaderboard | |
| 11 | Grok 4.1 Fast | 80.5% | Nov 2025 | Official leaderboard | |
| 12 | | 80% | Dec 2024 | Official leaderboard | |
| 13 | | 75.3% | Jun 2024 | Official leaderboard | |
| 14 | GPT-5 mini | 72.5% | Aug 2025 | Official leaderboard | |
| 15 | | 67% | Sep 2025 | Official leaderboard | |
| 16 | | 63.5% | Apr 2025 | Official leaderboard | |
| 17 | davinci-002 | 50% | Mar 2022 | Official leaderboard | |
| 18 | GPT-5 nano | 39.4% | Aug 2025 | Official leaderboard |
Rankings by part
We read 7 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Multiple Choice, Fill in the Blanks and Definition Writing. GPT-5.1 leads 3 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Multiple Choice
18 models we track have a score here.
- 1 GPT-5.1 100%
- 2
DeepSeek-V3.2 ↑5 100%
- 3
Mistral Large 3 ↑3 100%
- 4 GPT-5.2 ↓2 100%
- 5 GPT-3.5 Turbo Jan 2024 ↑3 96.7%
- 6 GPT-4o mini ↓3 96.7%
- 7 Grok 4.1 Fast ↑4 96.7%
- 8
Gemma 2 9B ↑5 93.3%
- 9
Llama 3.3 70B ↑3 93.3%
- 10
Gemma 3 12B 93.3%
Fill in the Blanks
18 models we track have a score here.
- 1 Grok 4 ↑3 93.3%
- 2 GPT-5.1 ↓1 90%
- 3 GPT-4o mini 86.7%
- 4 GPT-5.2 ↓2 86.7%
- 5 Claude Sonnet 4 80%
- 6 Claude 3 Haiku ↑3 70%
- 7
DeepSeek-V3.2 70%
- 8 GPT-5 mini ↑6 66.7%
- 9
Mistral Large 3 ↓3 63.3%
- 10 GPT-3.5 Turbo Jan 2024 ↓2 56.7%
Definition Writing
18 models we track have a score here.
- 1 GPT-5.1 100%
- 2 GPT-5.2 100%
- 3 GPT-4o mini 90%
- 4 Claude Sonnet 4 ↑1 90%
- 5
Mistral Large 3 ↑1 86.7%
- 6 Grok 4 ↓2 83.3%
- 7 Claude 3 Haiku ↑2 80%
- 8
DeepSeek-V3.2 ↓1 76.7%
- 9 GPT-3.5 Turbo Jan 2024 ↓1 73.3%
- 10
Gemma 2 9B ↑3 70%
Translation
18 models we track have a score here.
- 1 GPT-4o mini ↑2 100%
- 2 Claude Sonnet 4 ↑3 100%
- 3 GPT-5.1 ↓2 100%
- 4
Mistral Large 3 ↑2 100%
- 5 GPT-5.2 ↓3 100%
- 6 Claude 3 Haiku ↑3 95%
- 7
Llama 3.3 70B ↑5 95%
- 8 Grok 4 ↓4 95%
- 9 GPT-5 mini ↑5 95%
- 10 Grok 4.1 Fast ↑1 95%
Context
18 models we track have a score here.
- 1
Gemma 3 12B ↑9 96.7%
- 2 GPT-5.1 ↓1 96.7%
- 3 GPT-5.2 ↓1 96.7%
- 4 GPT-4o mini ↓1 93.3%
- 5 Grok 4 ↓1 93.3%
- 6
Llama 3.3 70B ↑6 90%
- 7 Claude Sonnet 4 ↓2 90%
- 8
Mistral Large 3 ↓2 90%
- 9 GPT-5 mini ↑5 86.7%
- 10 Claude 3 Haiku ↓1 83.3%
Match
18 models we track have a score here.
- 1 GPT-3.5 Turbo Jan 2024 ↑7 100%
- 2 Claude 3 Haiku ↑7 100%
- 3 GPT-4o mini 100%
- 4 Grok 4 100%
- 5 GPT-5.1 ↓4 100%
- 6 GPT-5.2 ↓4 96.7%
- 7
Gemma 3 12B ↑3 93.3%
- 8 Claude Sonnet 4 ↓3 93.3%
- 9
Llama 3.3 70B ↑3 90%
- 10
Mistral Large 3 ↓4 90%
Image
11 models we track have a score here.
- 1 GPT-5.1 96.7%
- 2 GPT-5.2 96.7%
- 3 Grok 4 ↑1 93.3%
- 4 Grok 4.1 Fast ↑7 93.3%
- 5 GPT-4o mini ↓2 86.7%
- 6 Claude Sonnet 4 ↓1 86.7%
- 7 GPT-5 mini ↑7 86.7%
- 8
Magistral Small 1.2 ↑7 86.7%
- 9
Mistral Large 3 ↓3 83.3%
- 10
Gemma 3 12B 80%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official LLM Benchmark: Danske Idiomer leaderboard
GPT-4o mini on the official leaderboard passed 85% in July 2024
Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official LLM Benchmark: Danske Idiomer 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 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (20 entries).
- Pass Models scored 18 models we track, plus 2 other entries on its leaderboard.
- Note Newest score The newest came 11 months ago, in December 2025, for GPT-5.2. That’s expected for a saturated test.
- Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.2 (#51).
- Note Headroom Saturated. The best is 97.5%, past 85% since July 2024.
Leaderboard import 3 pass
- Pass Importer We read the board on idiom.dk: 18 rows for models we track and 2 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.