tr_gender_bias
Models translate 200 Turkish passages into English. Set B scores pronoun accuracy on 82 passages where a name or kinship term identifies gender.
The author compares 24 translation systems, including GPT 5.5 and Gemini 3.1 Pro. Separate measures track occupational stereotypes and gender choices without a cue.
- Made by
- Boğaziçi University
- Released
- June 2026
- Kind
- Translation
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Propensity
- Behavior
- Bias
- Subject
- Languages and literature
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- GPT-5.5 #21 on the capability index
- Last model added
- May 2026
- Scores refreshed
- November 20th, 2026
3 models share the top score of 100% among models released by May 2026 3 models share the top score of 100% among models released by May 2026
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. 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 May 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 16 more
- 100.0 Cohere CAT+
- 100.0 GoogleTranslate
- 100.0 Lumen
- 100.0 SalamandraTA
- 100.0 TRIVE
- 100.0 VoxNexus_V1
- 100.0 Wayfinder
- 100.0
Gemma 4 31B IT
- 100.0 GPT-5.5
- 100.0
DeepSeek-V4-Pro
- 97.6 Gemma 4-4B
- 97.6 Qingqiu-MT-9B
- 95.1
Command A+
- 93.9 gpt-oss-120b
- 93.9
Qwen3.6 27B
- 92.7 Dubformer
- 91.5 gpt-oss-20b
- 90.2
Mistral Medium 3.5
- 87.8
Qwen3.5-9B
- 85.4
Tiny Aya Global
- 84.1 Tower 9B
- 84.1
Ministral 3 14B
- 81.7
Gemini 3.1 Pro
- 75.6 GLM-4-9B
- 1 Cohere CAT+ 100%
- 2 GoogleTranslate 100%
- 3 Lumen 100%
- 4 SalamandraTA 100%
- 5 TRIVE 100%
- 6 VoxNexus_V1 100%
- 7 Wayfinder 100%
- 8
Gemma 4 31B IT 100%
- 9 GPT-5.5 100%
- 10
DeepSeek-V4-Pro 100%
- 11 Gemma 4-4B 97.6%
- 12 Qingqiu-MT-9B 97.6%
- 13
Command A+ 95.1%
- 14 gpt-oss-120b 93.9%
- 15
Qwen3.6 27B 93.9%
- 16 Dubformer 92.7%
- 17 gpt-oss-20b 91.5%
- 18
Mistral Medium 3.5 90.2%
- 19
Qwen3.5-9B 87.8%
- 20
Tiny Aya Global 85.4%
- 21 Tower 9B 84.1%
- 22
Ministral 3 14B 84.1%
- 23
Gemini 3.1 Pro 81.7%
- 24 GLM-4-9B 75.6%
Source: The tr_gender_bias paper
- 1
Gemma 4 31B IT 100%
- 2 GPT-5.5 100%
- 3
DeepSeek-V4-Pro 100%
- 4
Command A+ 95.1%
- 5 gpt-oss-120b 93.9%
- 6
Qwen3.6 27B 93.9%
- 7 gpt-oss-20b 91.5%
- 8
Mistral Medium 3.5 90.2%
- 9
Qwen3.5-9B 87.8%
- 10
Tiny Aya Global 85.4%
- 11
Ministral 3 14B 84.1%
- 12
Gemini 3.1 Pro 81.7%
- 13 GLM-4-9B 75.6%
Source: The tr_gender_bias paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Cohere CAT+ | 100% | Official leaderboard | ||
| 2 | GoogleTranslate | 100% | Official leaderboard | ||
| 3 | Lumen | 100% | Official leaderboard | ||
| 4 | SalamandraTA | 100% | Official leaderboard | ||
| 5 | TRIVE | 100% | Official leaderboard | ||
| 6 | VoxNexus_V1 | 100% | Official leaderboard | ||
| 7 | Wayfinder | 100% | Official leaderboard | ||
| 8 | | 100% | Apr 2026 | Official leaderboard | |
| 9 | GPT-5.5 | 100% | Apr 2026 | Official leaderboard | |
| 10 | | 100% | Apr 2026 | Official leaderboard | |
| 11 | Gemma 4-4B | 97.6% | Official leaderboard | ||
| 12 | Qingqiu-MT-9B | 97.6% | Official leaderboard | ||
| 13 | | 95.1% | May 2026 | Official leaderboard | |
| 14 | gpt-oss-120b | 93.9% | Aug 2025 | Official leaderboard | |
| 15 | | 93.9% | Apr 2026 | Official leaderboard | |
| 16 | Dubformer | 92.7% | Official leaderboard | ||
| 17 | gpt-oss-20b | 91.5% | Aug 2025 | Official leaderboard | |
| 18 | | 90.2% | Apr 2026 | Official leaderboard | |
| 19 | | 87.8% | Feb 2026 | Official leaderboard | |
| 20 | | 85.4% | Feb 2026 | Official leaderboard | |
| 21 | Tower 9B | 84.1% | Official leaderboard | ||
| 22 | | 84.1% | Dec 2025 | Official leaderboard | |
| 23 | | 81.7% | Feb 2026 | Official leaderboard | |
| 24 | GLM-4-9B | 75.6% | Jun 2024 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 100% | Apr 2026 | Official leaderboard | |
| 2 | GPT-5.5 | 100% | Apr 2026 | Official leaderboard | |
| 3 | | 100% | Apr 2026 | Official leaderboard | |
| 4 | | 95.1% | May 2026 | Official leaderboard | |
| 5 | gpt-oss-120b | 93.9% | Aug 2025 | Official leaderboard | |
| 6 | | 93.9% | Apr 2026 | Official leaderboard | |
| 7 | gpt-oss-20b | 91.5% | Aug 2025 | Official leaderboard | |
| 8 | | 90.2% | Apr 2026 | Official leaderboard | |
| 9 | | 87.8% | Feb 2026 | Official leaderboard | |
| 10 | | 85.4% | Feb 2026 | Official leaderboard | |
| 11 | | 84.1% | Dec 2025 | Official leaderboard | |
| 12 | | 81.7% | Feb 2026 | Official leaderboard | |
| 13 | GLM-4-9B | 75.6% | Jun 2024 | Official leaderboard |
Rankings by part
We read 12 parts from the benchmark’s paper, each ranking models on one part of the test, such as Malformed, Hedge and Set A they. Gemini 3.1 Pro leads 3 of them, and 6 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Malformed
13 models we track have a score here.
- 1
Gemini 3.1 Pro ↑11 43%
- 2 GLM-4-9B ↑11 0%
- 3 gpt-oss-120b ↑2 0%
- 4 gpt-oss-20b ↑3 0%
- 5
Ministral 3 14B ↑6 0%
- 6
Tiny Aya Global ↑4 0%
- 7
Qwen3.5-9B ↑2 0%
- 8
Gemma 4 31B IT ↓7 0%
- 9
Qwen3.6 27B ↓3 0%
- 10 GPT-5.5 ↓8 0%
Hedge
13 models we track have a score here.
- 1 gpt-oss-120b ↑4 6.00%
- 2 gpt-oss-20b ↑5 4.00%
- 3
Gemini 3.1 Pro ↑9 3.50%
- 4 GLM-4-9B ↑9 2.00%
- 5
Gemma 4 31B IT ↓4 2.00%
- 6
Qwen3.6 27B 2.00%
- 7
Tiny Aya Global ↑3 1.50%
- 8 GPT-5.5 ↓6 1.00%
- 9
Ministral 3 14B ↑2 0.50%
- 10
Qwen3.5-9B ↓1 0.50%
Set A they
13 models we track have a score here.
- 1 GPT-5.5 ↑1 40.2%
- 2
Mistral Medium 3.5 ↑6 23.2%
- 3
DeepSeek-V4-Pro 19.5%
- 4
Ministral 3 14B ↑7 15.9%
- 5
Gemini 3.1 Pro ↑7 11%
- 6
Gemma 4 31B IT ↓5 9.80%
- 7 gpt-oss-20b 8.50%
- 8 GLM-4-9B ↑5 3.70%
- 9
Tiny Aya Global ↑1 2.40%
- 10
Qwen3.5-9B ↓1 2.40%
Set A stereotype agreement
13 models we track have a score here.
- 1
Gemini 3.1 Pro ↑11 100%
- 2 GPT-5.5 93.8%
- 3
Ministral 3 14B ↑8 89.2%
- 4
Tiny Aya Global ↑6 85.7%
- 5
Qwen3.6 27B ↑1 85.1%
- 6
Mistral Medium 3.5 ↑2 85%
- 7
Gemma 4 31B IT ↓6 83.3%
- 8
DeepSeek-V4-Pro ↓5 80%
- 9
Qwen3.5-9B 77.6%
- 10 GLM-4-9B ↑3 76%
Male-typed agreement
13 models we track have a score here.
- 1 GLM-4-9B ↑12 100%
- 2 gpt-oss-120b ↑3 100%
- 3 gpt-oss-20b ↑4 100%
- 4
Ministral 3 14B ↑7 100%
- 5
Tiny Aya Global ↑5 100%
- 6
Gemini 3.1 Pro ↑6 100%
- 7
Qwen3.5-9B ↑2 100%
- 8
Gemma 4 31B IT ↓7 100%
- 9
Qwen3.6 27B ↓3 100%
- 10 GPT-5.5 ↓8 100%
Female-typed agreement
13 models we track have a score here.
- 1
Gemini 3.1 Pro ↑11 100%
- 2 GPT-5.5 81.2%
- 3
Ministral 3 14B ↑8 77.4%
- 4
Tiny Aya Global ↑6 70.3%
- 5
Qwen3.6 27B ↑1 67.6%
- 6
Mistral Medium 3.5 ↑2 60.9%
- 7
Gemma 4 31B IT ↓6 60%
- 8
Qwen3.5-9B ↑1 51.6%
- 9
DeepSeek-V4-Pro ↓6 45.8%
- 10 GLM-4-9B ↑3 45.5%
Male-skew index
13 models we track have a score here.
- 1 gpt-oss-120b ↑4 0.76
- 2 gpt-oss-20b ↑5 0.76
- 3
Command A+ ↑1 0.76
- 4
Gemini 3.1 Pro ↑8 0.68
- 5
DeepSeek-V4-Pro ↓2 0.66
- 6 GLM-4-9B ↑7 0.60
- 7
Mistral Medium 3.5 ↑1 0.53
- 8
Qwen3.5-9B ↑1 0.52
- 9
Gemma 4 31B IT ↓8 0.50
- 10 GPT-5.5 ↓8 0.46
Set B accuracy
13 models we track have a score here.
- 1
Gemma 4 31B IT 100%
- 2 GPT-5.5 100%
- 3
DeepSeek-V4-Pro 100%
- 4
Command A+ 95.1%
- 5 gpt-oss-120b 93.9%
- 6
Qwen3.6 27B 93.9%
- 7 gpt-oss-20b 91.5%
- 8
Mistral Medium 3.5 90.2%
- 9
Qwen3.5-9B 87.8%
- 10
Tiny Aya Global 85.4%
ΔS
13 models we track have a score here.
- 1 GLM-4-9B ↑12 15.3%
- 2
Mistral Medium 3.5 ↑6 14.8%
- 3
Tiny Aya Global ↑7 11.4%
- 4 gpt-oss-20b ↑3 7.50%
- 5 gpt-oss-120b 3.80%
- 6
Qwen3.6 27B 3.80%
- 7
Ministral 3 14B ↑4 2.40%
- 8
Qwen3.5-9B ↑1 2.20%
- 9
Command A+ ↓5 2.00%
- 10
Gemini 3.1 Pro ↑2 0.70%
ΔG
13 models we track have a score here.
- 1
Ministral 3 14B ↑10 16.3%
- 2 GLM-4-9B ↑11 8.60%
- 3 gpt-oss-120b ↑2 7.00%
- 4
Qwen3.6 27B ↑2 7.00%
- 5 gpt-oss-20b ↑2 6.90%
- 6
Command A+ ↓2 4.60%
- 7
Qwen3.5-9B ↑2 4.30%
- 8
Gemma 4 31B IT ↓7 0%
- 9 GPT-5.5 ↓7 0%
- 10
DeepSeek-V4-Pro ↓7 0%
Set C they
13 models we track have a score here.
- 1 GPT-5.5 ↑1 88.9%
- 2
Ministral 3 14B ↑9 50%
- 3
Gemini 3.1 Pro ↑9 30.6%
- 4
DeepSeek-V4-Pro ↓1 27.8%
- 5
Mistral Medium 3.5 ↑3 27.8%
- 6
Gemma 4 31B IT ↓5 22.2%
- 7 GLM-4-9B ↑6 13.9%
- 8 gpt-oss-20b ↓1 8.30%
- 9
Qwen3.5-9B 8.30%
- 10
Qwen3.6 27B ↓4 8.30%
Set C over
13 models we track have a score here.
- 1
Tiny Aya Global ↑9 94.4%
- 2 GLM-4-9B ↑11 86.1%
- 3 gpt-oss-120b ↑2 83.3%
- 4 gpt-oss-20b ↑3 83.3%
- 5
Command A+ ↓1 77.8%
- 6
Qwen3.6 27B 75%
- 7
Qwen3.5-9B ↑2 72.2%
- 8
Gemma 4 31B IT ↓7 72.2%
- 9
DeepSeek-V4-Pro ↓6 72.2%
- 10
Mistral Medium 3.5 ↓2 58.3%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The tr_gender_bias paper
Gpt-oss-120b on the official leaderboard passed 85% in August 2025
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 tr_gender_bias 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 benchmark’s paper (24 entries).
- Pass Models scored 13 models we track, plus 11 other entries on its leaderboard.
- Note Newest score The newest came 6 months ago, in May 2026, for Command A+. 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-5.5 (#21).
- Note Headroom Saturated. The best is 100%, past 85% since August 2025.
Leaderboard import 3 pass
- Pass Importer We read the board on aclanthology.org: 13 rows for models we track and 11 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.