Benchmark Almanac

CoST English-Hindi

Claude-3.5 (snapshot unspecified) 81.5 tops its leaderboard Llama 3-70B 79.1 leads the models we track
Not in model cards

Models translate 1,947 English sentences into Hindi across eight genres, including plays, poetry and conversation. The organizers report BLEU alongside reference-based and reference-free translation metrics.

Passes test 4. The organizers evaluated 16 systems at WMT24 and reused the test for 22 at WMT26, including GPT-5.5 and Gemini 3.1 Pro.

Released
June 2024
Kind
Translation
Level
University
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English, Hindi
In and out
Text → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Llama 3-70B #182 on the capability index
Last model added
August 2024
Scores refreshed
November 20th, 2026
16
Entries in its paper, 2 models we track

Claude-3.5 (snapshot unspecified) tops the leaderboard with 81.5 Llama 3-70B leads with 79.1

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. The best model we track is Llama 3-70B, at 79.1. This view shows only models we track, each at its best.

  1. 1 Claude-3.5 (snapshot unspecified) 81.5
  2. 2 TranssionMT 81.5
  3. 3 ONLINE-B 81.4
  4. 4 Unbabel-Tower70B 80.9
  5. 5 Llama 3-70B 79.1
  6. 6 IOL_Research 79.0
  7. 7 GPT-4 (snapshot unspecified) 78.5
  8. 8 ONLINE-A 78.5
  9. 9 ZMT 78.5
  10. 10 Command R+ 78.5
  11. 11 Aya23 76.1
  12. 12 ONLINE-G 73.5
  13. 13 NVIDIA-NeMo 73.4
  14. 14 IKUN-C 65.8
  15. 15 IKUN 57.4
  16. 16 CycleL 36.6

Source: The CoST English-Hindi paper

Source: The CoST English-Hindi paper

#EntryScoreSettingDateSource
1 Claude-3.5 (snapshot unspecified) 81.5 Official leaderboard
2 TranssionMT 81.5 Official leaderboard
3 ONLINE-B 81.4 Official leaderboard
4 Unbabel-Tower70B 80.9 Official leaderboard
5 Llama 3-70B 79.1 Apr 2024 Official leaderboard
6 IOL_Research 79.0 Official leaderboard
7 GPT-4 (snapshot unspecified) 78.5 Official leaderboard
8 ONLINE-A 78.5 Official leaderboard
9 ZMT 78.5 Official leaderboard
10 Command R+ 78.5 Aug 2024 Official leaderboard
11 Aya23 76.1 Official leaderboard
12 ONLINE-G 73.5 Official leaderboard
13 NVIDIA-NeMo 73.4 Official leaderboard
14 IKUN-C 65.8 Official leaderboard
15 IKUN 57.4 Official leaderboard
16 CycleL 36.6 Official leaderboard
#ModelScoreSettingDateSource
1 Llama 3-70B 79.1 Apr 2024 Official leaderboard
2 Command R+ 78.5 Aug 2024 Official leaderboard

Scores over time

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

In the benchmark’s paper
78.4 78.6 78.8 79.0 79.2 2024 78.4 78.6 78.8 79.0 79.2 2024

Source: Official CoST English-Hindi 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

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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in August 2024, for Command R+. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (16 entries).
  • Pass Models scored 2 models we track, plus 14 other entries on its leaderboard.
  • Warning Newest score The newest came 2 years ago, in August 2024, for Command R+.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Llama 3-70B (#182).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on aclanthology.org: 2 rows for models we track and 14 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