Benchmark Almanac

IndicParam

Gemini-2.5-flash (snapshot unspecified) 58% tops its leaderboard GPT-5 45% leads the models we track
Not in model cards

Models answer multiple-choice questions from Indian competitive exams. The benchmark tests knowledge in low-resource Indic languages.

Released
November 2025
Kind
Knowledge in many languages
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
Many languages
In and out
Text → Text
Interaction
Single turn
Items from
Exams and contests
Refresh
One fixed set
Run by
Anyone
In views
Many languages
Strongest model tested
GPT-5 #71 on the capability index
Last model added
October 2025
Scores refreshed
November 20th, 2026
20
Entries in its paper, 19 models we track

Gemini-2.5-flash (snapshot unspecified) tops the leaderboard with 58% GPT-5 leads with 45% among models released by October 2025

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 GPT-5, at 45%. This view shows only models we track, each at its best.

The newest model with a score here came out in October 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 Gemini-2.5-flash (snapshot unspecified) Zero-shot 58%
  2. 2 GPT-5 Zero-shot 45%
  3. 3 DeepSeek-V3.2-Exp Zero-shot 43.1%
  4. 4 Claude Haiku 4.5 Zero-shot 42.7%
  5. 5 Grok 4 Fast Zero-shot 39.6%
  6. 6 Llama 3.3 70B Zero-shot 38.6%
  7. 7 Qwen3-32B Zero-shot 37.7%
  8. 8 gpt-oss-120b Zero-shot, low reasoning 37.7%
  9. 9 Gemma 3 27B Zero-shot 37.3%
  10. 10 Llama 4 Scout Zero-shot 35.5%
  11. 11 Gemma 3 12B Zero-shot 35.3%
  12. 12 Mistral Small 3.1 Zero-shot 35.1%
  13. 13 Qwen3-30B-A3B Zero-shot, thinking disabled 33.5%
  14. 14 Aya Expanse 32B Zero-shot 33.1%
  15. 15 qwen3-4b-instruct-2507 Zero-shot 30.8%
  16. 16 Qwen2.5-3B Zero-shot 30.1%
  17. 17 Llama 3.1-8B Zero-shot 29.6%
  18. 18 Aya Expanse 8B Zero-shot 29.1%
  19. 19 Gemma 3 4B Zero-shot 28.6%
  20. 20 Llama 3.2 3B Zero-shot 27%

Source: The IndicParam paper

#EntryScoreSettingDateSource
1 Gemini-2.5-flash (snapshot unspecified) 58% Zero-shot Official leaderboard
2 GPT-5 45% Zero-shot Aug 2025 Official leaderboard
3 DeepSeek-V3.2-Exp 43.1% Zero-shot Sep 2025 Official leaderboard
4 Claude Haiku 4.5 42.7% Zero-shot Oct 2025 Official leaderboard
5 Grok 4 Fast 39.6% Zero-shot Sep 2025 Official leaderboard
6 Llama 3.3 70B 38.6% Zero-shot Dec 2024 Official leaderboard
7 Qwen3-32B 37.7% Zero-shot Apr 2025 Official leaderboard
8 gpt-oss-120b 37.7% Zero-shot, low reasoning Aug 2025 Official leaderboard
9 Gemma 3 27B 37.3% Zero-shot Mar 2025 Official leaderboard
10 Llama 4 Scout 35.5% Zero-shot Apr 2025 Official leaderboard
11 Gemma 3 12B 35.3% Zero-shot Mar 2025 Official leaderboard
12 Mistral Small 3.1 35.1% Zero-shot Mar 2025 Official leaderboard
13 Qwen3-30B-A3B 33.5% Zero-shot, thinking disabled Apr 2025 Official leaderboard
14 Aya Expanse 32B 33.1% Zero-shot Oct 2024 Official leaderboard
15 qwen3-4b-instruct-2507 30.8% Zero-shot Aug 2025 Official leaderboard
16 Qwen2.5-3B 30.1% Zero-shot Sep 2024 Official leaderboard
17 Llama 3.1-8B 29.6% Zero-shot Jul 2024 Official leaderboard
18 Aya Expanse 8B 29.1% Zero-shot Oct 2024 Official leaderboard
19 Gemma 3 4B 28.6% Zero-shot Mar 2025 Official leaderboard
20 Llama 3.2 3B 27% Zero-shot Sep 2024 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5 45% Zero-shot Aug 2025 Official leaderboard
2 DeepSeek-V3.2-Exp 43.1% Zero-shot Sep 2025 Official leaderboard
3 Claude Haiku 4.5 42.7% Zero-shot Oct 2025 Official leaderboard
4 Grok 4 Fast 39.6% Zero-shot Sep 2025 Official leaderboard
5 Llama 3.3 70B 38.6% Zero-shot Dec 2024 Official leaderboard
6 Qwen3-32B 37.7% Zero-shot Apr 2025 Official leaderboard
7 gpt-oss-120b 37.7% Zero-shot, low reasoning Aug 2025 Official leaderboard
8 Gemma 3 27B 37.3% Zero-shot Mar 2025 Official leaderboard
9 Llama 4 Scout 35.5% Zero-shot Apr 2025 Official leaderboard
10 Gemma 3 12B 35.3% Zero-shot Mar 2025 Official leaderboard
11 Mistral Small 3.1 35.1% Zero-shot Mar 2025 Official leaderboard
12 Qwen3-30B-A3B 33.5% Zero-shot, thinking disabled Apr 2025 Official leaderboard
13 Aya Expanse 32B 33.1% Zero-shot Oct 2024 Official leaderboard
14 qwen3-4b-instruct-2507 30.8% Zero-shot Aug 2025 Official leaderboard
15 Qwen2.5-3B 30.1% Zero-shot Sep 2024 Official leaderboard
16 Llama 3.1-8B 29.6% Zero-shot Jul 2024 Official leaderboard
17 Aya Expanse 8B 29.1% Zero-shot Oct 2024 Official leaderboard
18 Gemma 3 4B 28.6% Zero-shot Mar 2025 Official leaderboard
19 Llama 3.2 3B 27% Zero-shot Sep 2024 Official leaderboard

Rankings by part

We read 42 parts from the benchmark’s paper, each ranking models on one part of the test, such as Dogri, Maithili and Rajasthani. GPT-5 leads 27 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Dogri

19 models we track have a score here.

  1. 1 GPT-5 45.7%
  2. 2 Claude Haiku 4.5 ↑1 44.8%
  3. 3 Qwen3-32B ↑3 43.4%
  4. 4 DeepSeek-V3.2-Exp ↓2 43.3%
  5. 5 Grok 4 Fast ↓1 40.2%
  6. 6 gpt-oss-120b ↑1 39.7%
  7. 7 Mistral Small 3.1 ↑4 38%
  8. 8 Gemma 3 27B 37.6%
  9. 9 Llama 3.3 70B ↓4 37.3%
  10. 10 Gemma 3 12B 37.2%

Number one over time

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

OpenAI
Benchmark paper Not tested since Nov 2025
GPT-5 45%

Source: The IndicParam paper

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
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-5 Llama 3.3 70B Aya Expanse 32B Llama 3.1-8B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-5 Llama 3.3 70B

Source: Official IndicParam 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

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 13 months ago, in October 2025, for Claude Haiku 4.5. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (20 entries).
  • Pass Models scored 19 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 13 months ago, in October 2025, for Claude Haiku 4.5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
  • Pass Headroom The best is 45%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on aclanthology.org: 19 rows for models we track and 1 other entry 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