Benchmark Almanac

CLEF JOKER 2025 Task 2: Wordplay Translation

dsgt_o4_mini_multi_agent_discriminator 74% tops its leaderboard
Not in model cards
Version 2025

Models translate English puns into French. A translator judges whether each sampled answer preserves the meaning and contains wordplay. The online leaderboard uses BLEU.

Nine teams submitted 52 runs on 1,682 test puns. Georgia Tech submitted two o4-mini systems and another team fine-tuned Qwen models.

Released
October 2024
Kind
Translation
Level
University
Scored
Graded by people
Answers
Open-ended
Tasks
Public, plus a private test set
Measures
Capability
Language
English, French
In and out
Text → Text
Interaction
Single turn
Items from
Translated
Run by
A host
44
Entries in its paper, 0 models we track

dsgt_o4_mini_multi_agent_discriminator tops the leaderboard with 74%

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.

44 entries. Scroll sideways for the rest.

  1. 74.0 dsgt_o4_mini_multi_agent_discriminator
  2. 72.0 dsgt_o4_mini_chain_of_thought_phonetic_embeddings
  3. 52.0 duth_xanthi_argos
  4. 50.0 Cryptix
  5. 50.0 duth_google_flant5_fallback
  6. 50.0 duth_xanthi_GoogleTranslate
  7. 50.0 duth_xanthi_GoogleTranslate_fallback
  8. 48.0 UvA_finetunedNLLB-1.3B
  9. 48.0 UvA_finetunedNLLB-1.3B&finetunedroBERTa
  10. 46.0 pjmathematician_Q25-14
  11. 46.0 UvA_finetunedT5-base
  12. 46.0 UvA_T5-base&finetunedroBERTa
  13. 44.0 Skommarkhos_Lucie_SFT_ARPO
  14. 44.0 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_v8
  15. 44.0 UvA_finetunedMarianMT
  16. 44.0 UvA_finetunedMarianMT&finetunedroBERTa
  17. 42.0 Skommarkhos_Lucie_SFT
  18. 42.0 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_v4
  19. 42.0 teamX_aug
  20. 40.0 dsgt_simple_mistral_medium
  21. 40.0 Skommarkhos_Lucie-7B-Instruct-v1.1
  22. 40.0 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_arpo_a19
  23. 40.0 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a1
  24. 40.0 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a11
  25. 40.0 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a5
  26. 40.0 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a7
  27. 38.0 Cryptix_marianmt
  28. 38.0 duth_hybrid_fusion
  29. 38.0 duth_xanthi_helsinki
  30. 38.0 yourteamid_marianmt_pun_postedit
  31. 36.0 duth_xanthi_m2m100_1_2B
  32. 36.0 Skommarkhos_Croissant_SFT_ARPO
  33. 34.0 duth_xanthi_t5_base_gpu
  34. 34.0 Skommarkhos_Croissant_SFT
  35. 32.0 duth_combined_m2m100
  36. 30.0 UvA_finetunedmBARTcc25
  37. 28.0 duth_xanthi_mbart50
  38. 26.0 teamX_final
  39. 24.0 dsgt_simple_o4_mini
  40. 22.0 duth_xanthi_bloomz3b_local
  41. 22.0 UvA_mBARTcc25&finetunedroBERTa
  42. 4.00 Cryptix_finetunedmarian
  43. 2.00 Cryptix_rulebased
  44. 2.00 yourteam_rulebased
  1. 1 dsgt_o4_mini_multi_agent_discriminator 74%
  2. 2 dsgt_o4_mini_chain_of_thought_phonetic_embeddings 72%
  3. 3 duth_xanthi_argos 52%
  4. 4 Cryptix 50%
  5. 5 duth_google_flant5_fallback 50%
  6. 6 duth_xanthi_GoogleTranslate 50%
  7. 7 duth_xanthi_GoogleTranslate_fallback 50%
  8. 8 UvA_finetunedNLLB-1.3B 48%
  9. 9 UvA_finetunedNLLB-1.3B&finetunedroBERTa 48%
  10. 10 pjmathematician_Q25-14 46%
  11. 11 UvA_finetunedT5-base 46%
  12. 12 UvA_T5-base&finetunedroBERTa 46%
  13. 13 Skommarkhos_Lucie_SFT_ARPO 44%
  14. 14 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_v8 44%
  15. 15 UvA_finetunedMarianMT 44%
  16. 16 UvA_finetunedMarianMT&finetunedroBERTa 44%
  17. 17 Skommarkhos_Lucie_SFT 42%
  18. 18 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_v4 42%
  19. 19 teamX_aug 42%
  20. 20 dsgt_simple_mistral_medium 40%
  21. 21 Skommarkhos_Lucie-7B-Instruct-v1.1 40%
  22. 22 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_arpo_a19 40%
  23. 23 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a1 40%
  24. 24 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a11 40%
  25. 25 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a5 40%
  26. 26 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a7 40%
  27. 27 Cryptix_marianmt 38%
  28. 28 duth_hybrid_fusion 38%
  29. 29 duth_xanthi_helsinki 38%
  30. 30 yourteamid_marianmt_pun_postedit 38%
  31. 31 duth_xanthi_m2m100_1_2B 36%
  32. 32 Skommarkhos_Croissant_SFT_ARPO 36%
  33. 33 duth_xanthi_t5_base_gpu 34%
  34. 34 Skommarkhos_Croissant_SFT 34%
  35. 35 duth_combined_m2m100 32%
  36. 36 UvA_finetunedmBARTcc25 30%
  37. 37 duth_xanthi_mbart50 28%
  38. 38 teamX_final 26%
  39. 39 dsgt_simple_o4_mini 24%
  40. 40 duth_xanthi_bloomz3b_local 22%
  41. 41 UvA_mBARTcc25&finetunedroBERTa 22%
  42. 42 Cryptix_finetunedmarian 4.00%
  43. 43 Cryptix_rulebased 2.00%
  44. 44 yourteam_rulebased 2.00%

Source: The CLEF JOKER 2025 Task 2: Wordplay Translation paper

#EntryScoreSettingDateSource
1 dsgt_o4_mini_multi_agent_discriminator 74% Official leaderboard
2 dsgt_o4_mini_chain_of_thought_phonetic_embeddings 72% Official leaderboard
3 duth_xanthi_argos 52% Official leaderboard
4 Cryptix 50% Official leaderboard
5 duth_google_flant5_fallback 50% Official leaderboard
6 duth_xanthi_GoogleTranslate 50% Official leaderboard
7 duth_xanthi_GoogleTranslate_fallback 50% Official leaderboard
8 UvA_finetunedNLLB-1.3B 48% Official leaderboard
9 UvA_finetunedNLLB-1.3B&finetunedroBERTa 48% Official leaderboard
10 pjmathematician_Q25-14 46% Official leaderboard
11 UvA_finetunedT5-base 46% Official leaderboard
12 UvA_T5-base&finetunedroBERTa 46% Official leaderboard
13 Skommarkhos_Lucie_SFT_ARPO 44% Official leaderboard
14 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_v8 44% Official leaderboard
15 UvA_finetunedMarianMT 44% Official leaderboard
16 UvA_finetunedMarianMT&finetunedroBERTa 44% Official leaderboard
17 Skommarkhos_Lucie_SFT 42% Official leaderboard
18 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_v4 42% Official leaderboard
19 teamX_aug 42% Official leaderboard
20 dsgt_simple_mistral_medium 40% Official leaderboard
21 Skommarkhos_Lucie-7B-Instruct-v1.1 40% Official leaderboard
22 Skommarkhos_skommarkhos_lucie7binstructv1_1_sft_arpo_a19 40% Official leaderboard
23 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a1 40% Official leaderboard
24 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a11 40% Official leaderboard
25 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a5 40% Official leaderboard
26 Skommarkhos_skommarkhos-lucie7binstructv1-1-sft-arpo-a7 40% Official leaderboard
27 Cryptix_marianmt 38% Official leaderboard
28 duth_hybrid_fusion 38% Official leaderboard
29 duth_xanthi_helsinki 38% Official leaderboard
30 yourteamid_marianmt_pun_postedit 38% Official leaderboard
31 duth_xanthi_m2m100_1_2B 36% Official leaderboard
32 Skommarkhos_Croissant_SFT_ARPO 36% Official leaderboard
33 duth_xanthi_t5_base_gpu 34% Official leaderboard
34 Skommarkhos_Croissant_SFT 34% Official leaderboard
35 duth_combined_m2m100 32% Official leaderboard
36 UvA_finetunedmBARTcc25 30% Official leaderboard
37 duth_xanthi_mbart50 28% Official leaderboard
38 teamX_final 26% Official leaderboard
39 dsgt_simple_o4_mini 24% Official leaderboard
40 duth_xanthi_bloomz3b_local 22% Official leaderboard
41 UvA_mBARTcc25&finetunedroBERTa 22% Official leaderboard
42 Cryptix_finetunedmarian 4.00% Official leaderboard
43 Cryptix_rulebased 2.00% Official leaderboard
44 yourteam_rulebased 2.00% Official 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 8 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 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (44 entries).
  • Pass Models scored 44 entries on its leaderboard, none a model we track.
  • Note Newest score No scores yet.
  • Note Frontier models No models scored yet.
  • Note Headroom No scores yet.
Leaderboard import 3 pass
  • Pass Importer We read the board on ceur-ws.org: 44 entries at the last read, none a model we track.
  • 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