Benchmark Almanac

CAPR-ar

Gemma 3 27B 70.1% leads since July 2026
Not in model cards

Models read an Arabic scenario and choose the best matching proverb from four options. The authors score exact choices.

The makers rank 14 models on 1,222 scenarios. Every test target proverb also appears as a distractor in the training choices.

Released
July 2026
Kind
Reasoning in many languages
Level
Everyday
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
Arabic
In and out
Text → Text
Interaction
Single turn
Run by
Anyone
Strongest model tested
Qwen3-32B #135 on the capability index
Last model added
August 2025
Scores refreshed
November 20th, 2026
14
Entries on its own leaderboard, 12 models we track

Gemma 3 27B leads with 70.1% among models released by August 2025 Gemma 3 27B leads with 70.1% among models released by August 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 August 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemma 3 27B zero_shot 70.1%
  2. 2 Qwen3-32B zero_shot 67.4%
  3. 3 Qwen3-14B zero_shot 66.5%
  4. 4 Qwen3-8B zero_shot 65.5%
  5. 5 Gemma 3 12B zero_shot 65.3%
  6. 6 Aya Expanse 32B zero_shot 63.3%
  7. 7 Qwen3-4B zero_shot 58.1%
  8. 8 aya:35b zero_shot 54.5%
  9. 9 Aya Expanse 8B zero_shot 53.5%
  10. 10 Gemma 3 4B zero_shot 50.6%
  11. 11 Qwen3-1.7B zero_shot 44.7%
  12. 12 aya:8b zero_shot 44.4%
  13. 13 Gemma 3 1B zero_shot 30%
  14. 14 Gemma 3 270M zero_shot 24.5%

Source: Official CAPR-ar leaderboard

#EntryScoreSettingDateSource
1 Gemma 3 27B 70.1% zero_shot Mar 2025 Official leaderboard
2 Qwen3-32B 67.4% zero_shot Apr 2025 Official leaderboard
3 Qwen3-14B 66.5% zero_shot Apr 2025 Official leaderboard
4 Qwen3-8B 65.5% zero_shot Apr 2025 Official leaderboard
5 Gemma 3 12B 65.3% zero_shot Mar 2025 Official leaderboard
6 Aya Expanse 32B 63.3% zero_shot Oct 2024 Official leaderboard
7 Qwen3-4B 58.1% zero_shot Apr 2025 Official leaderboard
8 aya:35b 54.5% zero_shot Official leaderboard
9 Aya Expanse 8B 53.5% zero_shot Oct 2024 Official leaderboard
10 Gemma 3 4B 50.6% zero_shot Mar 2025 Official leaderboard
11 Qwen3-1.7B 44.7% zero_shot Apr 2025 Official leaderboard
12 aya:8b 44.4% zero_shot Official leaderboard
13 Gemma 3 1B 30% zero_shot Mar 2025 Official leaderboard
14 Gemma 3 270M 24.5% zero_shot Aug 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Gemma 3 27B 70.1% zero_shot Mar 2025 Official leaderboard
2 Qwen3-32B 67.4% zero_shot Apr 2025 Official leaderboard
3 Qwen3-14B 66.5% zero_shot Apr 2025 Official leaderboard
4 Qwen3-8B 65.5% zero_shot Apr 2025 Official leaderboard
5 Gemma 3 12B 65.3% zero_shot Mar 2025 Official leaderboard
6 Aya Expanse 32B 63.3% zero_shot Oct 2024 Official leaderboard
7 Qwen3-4B 58.1% zero_shot Apr 2025 Official leaderboard
8 Aya Expanse 8B 53.5% zero_shot Oct 2024 Official leaderboard
9 Gemma 3 4B 50.6% zero_shot Mar 2025 Official leaderboard
10 Qwen3-1.7B 44.7% zero_shot Apr 2025 Official leaderboard
11 Gemma 3 1B 30% zero_shot Mar 2025 Official leaderboard
12 Gemma 3 270M 24.5% zero_shot Aug 2025 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Own leaderboard Not tested since Jul 2026
Gemma 3 27B 70.1%

Source: Official CAPR-ar leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Gemma 3 27B Aya Expanse 32B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Gemma 3 27B Aya Expanse 32B

Source: Official CAPR-ar 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.

Versions

More in Multilingual

All 545 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 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (14 entries).
  • Pass Models scored 12 models we track, plus 2 other entries on its leaderboard.
  • Note Newest score The newest came 15 months ago, in August 2025, for Gemma 3 270M. That’s expected for a test replaced by CAPR-am.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Qwen3-32B (#135).
  • Pass Headroom The best is 70.1%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 12 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.

How we check