Benchmark Almanac

INCLUDE

Claude Opus 5 89.8% leads since July 2026
Aya Expanse 32B 59.5% leads since November 2024
Claude Opus 5 89.8% leads since July 2026
Fading

Models answer multiple-choice questions from local academic and professional exams in 44 languages that test regional knowledge. Scored by accuracy.

INCLUDE-base has 22,637 exam questions in 44 languages. A lite subset of 11,095 questions also exists, and the dataset cards show scores within about a point of the base set. Reports rarely say which set they used.

Made by
EPFL and Cohere For AI
Released
November 2024
Kind
Knowledge in many languages
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Strongest model tested
Claude Opus 5 #10 on the capability index
Last in a launch post
June 2026, dropped after it saturated
Last model added
July 2026
Scores refreshed
November 14th, 2026

In the authors’ words

The performance differential of large language models (LLM) between languages hinders their effective deployment in many regions, inhibiting the potential economic and societal value of generative AI tools in many communities. However, the development of functional LLMs in many languages (i.e., multilingual LLMs) is bottlenecked by the lack of high-quality evaluation resources in languages other than English. Moreover, current practices in multilingual benchmark construction often translate English resources, ignoring the regional and cultural knowledge of the environments in which multilingual systems would be used. In this work, we construct an evaluation suite of 197,243 QA pairs from local exam sources to measure the capabilities of multilingual LLMs in a variety of regional contexts. Our novel resource, INCLUDE, is a comprehensive knowledge- and reasoning-centric benchmark across 44 written languages that evaluates multilingual LLMs for performance in the actual language environments where they would be deployed.

Abstract of INCLUDE: Evaluating Multilingual Language Understanding with Regional Knowledge (arXiv:2411.19799, CC0)
31
Models scored in lab cards
19
Launch posts list it, from 3 labs
14
System cards report it, from 4 labs

Claude Opus 5 leads with 89.8% among models released by July 2026

Best score per model as labs report them in their own cards, drawn striped. Labs pick their own settings, so these compare less cleanly than independent runs.

Aya Expanse 32B leads with 59.5% across all independent runs

Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Dotted bars are reruns by Epoch AI or another board, and each row names its source.

Claude Opus 5 leads with 89.8%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

The newest model with a score here came out in July 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

31 models. Scroll sideways for the rest.

  1. 89.8 Claude Opus 5
  2. 87.6 Claude Opus 4.8
  3. 87.0 Claude Opus 4.7
  4. 86.5 Claude Sonnet 5
  5. 86.2 Qwen3.7-Max
  6. 85.6 Qwen3.5 397B-A17B
  7. 85.1 Qwen 3.6 Plus
  8. 83.0 Qwen3.7-Plus
  9. 82.8 Qwen3.5-122B-A10B
  10. 81.6 Qwen3.5-27B
  11. 81.0 Qwen3-235B-A22B-Thinking
  12. 80.0 Qwen3-VL-235B-A22B
  13. 79.7 Qwen3.5-35B-A3B
  14. 78.9 Qwen3-Next-80B-A3B
  15. 78.7 Qwen3-235B-A22B
  16. 75.6 Qwen3.5-9B
  17. 74.3 Cohere Command A
  18. 73.7 Qwen3-32B
  19. 71.9 Qwen3-30B-A3B
  20. 71.7 Qwen3-14B
  21. 71.0 Qwen3.5-4B
  22. 67.8 Qwen3-8B
  23. 67.3 Granite 4.1 30B
  24. 64.0 Granite 4.0 H Small
  25. 61.8 Qwen3-4B
  26. 58.2 Apertus 70B
  27. 57.2 Gemma 3n E4B
  28. 55.4 Qwen3.5-2B
  29. 51.8 Qwen3-1.7B
  30. 42.1 granite-4.0-1b
  31. 40.6 Qwen3.5-0.8B
  1. 1 Claude Opus 5 max effort 89.8%
  2. 2 Claude Opus 4.8 87.6%
  3. 3 Claude Opus 4.7 87%
  4. 4 Claude Sonnet 5 max effort 86.5%
  5. 5 Qwen3.7-Max 86.2%
  6. 6 Qwen3.5 397B-A17B 85.6%
  7. 7 Qwen 3.6 Plus 85.1%
  8. 8 Qwen3.7-Plus 83%
  9. 9 Qwen3.5-122B-A10B 82.8%
  10. 10 Qwen3.5-27B 81.6%
  11. 11 Qwen3-235B-A22B-Thinking Jul 2025 81%
  12. 12 Qwen3-VL-235B-A22B 80%
  13. 13 Qwen3.5-35B-A3B 79.7%
  14. 14 Qwen3-Next-80B-A3B 78.9%
  15. 15 Qwen3-235B-A22B 78.7%
  16. 16 Qwen3.5-9B 75.6%
  17. 17 Cohere Command A 74.3%
  18. 18 Qwen3-32B 73.7%
  19. 19 Qwen3-30B-A3B 71.9%
  20. 20 Qwen3-14B 71.7%
  21. 21 Qwen3.5-4B 71%
  22. 22 Qwen3-8B 67.8%
  23. 23 Granite 4.1 30B 67.3%
  24. 24 Granite 4.0 H Small 64%
  25. 25 Qwen3-4B 61.8%
  26. 26 Apertus 70B 58.2%
  27. 27 Gemma 3n E4B 57.2%
  28. 28 Qwen3.5-2B 55.4%
  29. 29 Qwen3-1.7B 51.8%
  30. 30 granite-4.0-1b 42.1%
  31. 31 Qwen3.5-0.8B 40.6%

Source: Labs’ model cards

Source: Official INCLUDE leaderboard

34 models. Scroll sideways for the rest.

  1. 89.8 Claude Opus 5
  2. 87.6 Claude Opus 4.8
  3. 87.0 Claude Opus 4.7
  4. 86.5 Claude Sonnet 5
  5. 86.2 Qwen3.7-Max
  6. 85.6 Qwen3.5 397B-A17B
  7. 85.1 Qwen 3.6 Plus
  8. 83.0 Qwen3.7-Plus
  9. 82.8 Qwen3.5-122B-A10B
  10. 81.6 Qwen3.5-27B
  11. 81.0 Qwen3-235B-A22B-Thinking
  12. 80.0 Qwen3-VL-235B-A22B
  13. 79.7 Qwen3.5-35B-A3B
  14. 78.9 Qwen3-Next-80B-A3B
  15. 78.7 Qwen3-235B-A22B
  16. 75.6 Qwen3.5-9B
  17. 74.3 Cohere Command A
  18. 73.7 Qwen3-32B
  19. 71.9 Qwen3-30B-A3B
  20. 71.7 Qwen3-14B
  21. 71.0 Qwen3.5-4B
  22. 67.8 Qwen3-8B
  23. 67.3 Granite 4.1 30B
  24. 64.0 Granite 4.0 H Small
  25. 61.8 Qwen3-4B
  26. 59.5 Aya Expanse 32B
  27. 58.2 Apertus 70B
  28. 57.2 Gemma 3n E4B
  29. 55.4 Qwen3.5-2B
  30. 51.8 Qwen3-1.7B
  31. 47.8 Aya Expanse 8B
  32. 44.3 Mistral-7B-Instruct-v0.1
  33. 42.1 granite-4.0-1b
  34. 40.6 Qwen3.5-0.8B
  1. 1 Claude Opus 5 max effort 89.8%
  2. 2 Claude Opus 4.8 87.6%
  3. 3 Claude Opus 4.7 87%
  4. 4 Claude Sonnet 5 max effort 86.5%
  5. 5 Qwen3.7-Max 86.2%
  6. 6 Qwen3.5 397B-A17B 85.6%
  7. 7 Qwen 3.6 Plus 85.1%
  8. 8 Qwen3.7-Plus 83%
  9. 9 Qwen3.5-122B-A10B 82.8%
  10. 10 Qwen3.5-27B 81.6%
  11. 11 Qwen3-235B-A22B-Thinking Jul 2025 81%
  12. 12 Qwen3-VL-235B-A22B 80%
  13. 13 Qwen3.5-35B-A3B 79.7%
  14. 14 Qwen3-Next-80B-A3B 78.9%
  15. 15 Qwen3-235B-A22B 78.7%
  16. 16 Qwen3.5-9B 75.6%
  17. 17 Cohere Command A 74.3%
  18. 18 Qwen3-32B 73.7%
  19. 19 Qwen3-30B-A3B 71.9%
  20. 20 Qwen3-14B 71.7%
  21. 21 Qwen3.5-4B 71%
  22. 22 Qwen3-8B 67.8%
  23. 23 Granite 4.1 30B 67.3%
  24. 24 Granite 4.0 H Small 64%
  25. 25 Qwen3-4B 61.8%
  26. 26 Aya Expanse 32B English instructions 59.5%
  27. 27 Apertus 70B 58.2%
  28. 28 Gemma 3n E4B 57.2%
  29. 29 Qwen3.5-2B 55.4%
  30. 30 Qwen3-1.7B 51.8%
  31. 31 Aya Expanse 8B English instructions 47.8%
  32. 32 Mistral-7B-Instruct-v0.1 English instructions 44.3%
  33. 33 granite-4.0-1b 42.1%
  34. 34 Qwen3.5-0.8B 40.6%

Sources: Labs’ model cards and official INCLUDE leaderboard

#ModelScoreSettingDateSource
1 Claude Opus 5 89.8% max effort Jul 2026 Lab model card
2 Claude Opus 4.8 87.6% May 2026 Lab model card
3 Claude Opus 4.7 87% Apr 2026 Lab model card
4 Claude Sonnet 5 86.5% max effort Jun 2026 Lab model card
5 Qwen3.7-Max 86.2% May 2026 Lab model card
6 Qwen3.5 397B-A17B 85.6% Feb 2026 Lab model card
7 Qwen 3.6 Plus 85.1% Apr 2026 Lab model card
8 Qwen3.7-Plus 83% Jun 2026 Lab model card
9 Qwen3.5-122B-A10B 82.8% Feb 2026 Lab model card
10 Qwen3.5-27B 81.6% Feb 2026 Lab model card
11 Qwen3-235B-A22B-Thinking Jul 2025 81% Jul 2025 Lab model card
12 Qwen3-VL-235B-A22B 80% Sep 2025 Lab model card
13 Qwen3.5-35B-A3B 79.7% Feb 2026 Lab model card
14 Qwen3-Next-80B-A3B 78.9% Sep 2025 Lab model card
15 Qwen3-235B-A22B 78.7% Apr 2025 Lab model card
16 Qwen3.5-9B 75.6% Mar 2026 Lab model card
17 Cohere Command A 74.3% Apr 2025 Lab model card
18 Qwen3-32B 73.7% May 2025 Lab model card
19 Qwen3-30B-A3B 71.9% May 2025 Lab model card
20 Qwen3-14B 71.7% May 2025 Lab model card
21 Qwen3.5-4B 71% Mar 2026 Lab model card
22 Qwen3-8B 67.8% May 2025 Lab model card
23 Granite 4.1 30B 67.3% Apr 2026 Lab model card
24 Granite 4.0 H Small 64% Oct 2025 Lab model card
25 Qwen3-4B 61.8% May 2025 Lab model card
26 Apertus 70B 58.2% Sep 2025 Lab model card
27 Gemma 3n E4B 57.2% Jun 2025 Lab model card
28 Qwen3.5-2B 55.4% Mar 2026 Lab model card
29 Qwen3-1.7B 51.8% May 2025 Lab model card
30 granite-4.0-1b 42.1% Oct 2025 Lab model card
31 Qwen3.5-0.8B 40.6% Mar 2026 Lab model card
#ModelScoreSettingDateSource
1 Aya Expanse 32B 59.5% English instructions Oct 2024 Official leaderboard
2 Aya Expanse 8B 47.8% English instructions Oct 2024 Official leaderboard
3 Mistral-7B-Instruct-v0.1 44.3% English instructions Sep 2023 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 5 89.8% max effort Jul 2026 Lab model card
2 Claude Opus 4.8 87.6% May 2026 Lab model card
3 Claude Opus 4.7 87% Apr 2026 Lab model card
4 Claude Sonnet 5 86.5% max effort Jun 2026 Lab model card
5 Qwen3.7-Max 86.2% May 2026 Lab model card
6 Qwen3.5 397B-A17B 85.6% Feb 2026 Lab model card
7 Qwen 3.6 Plus 85.1% Apr 2026 Lab model card
8 Qwen3.7-Plus 83% Jun 2026 Lab model card
9 Qwen3.5-122B-A10B 82.8% Feb 2026 Lab model card
10 Qwen3.5-27B 81.6% Feb 2026 Lab model card
11 Qwen3-235B-A22B-Thinking Jul 2025 81% Jul 2025 Lab model card
12 Qwen3-VL-235B-A22B 80% Sep 2025 Lab model card
13 Qwen3.5-35B-A3B 79.7% Feb 2026 Lab model card
14 Qwen3-Next-80B-A3B 78.9% Sep 2025 Lab model card
15 Qwen3-235B-A22B 78.7% Apr 2025 Lab model card
16 Qwen3.5-9B 75.6% Mar 2026 Lab model card
17 Cohere Command A 74.3% Apr 2025 Lab model card
18 Qwen3-32B 73.7% May 2025 Lab model card
19 Qwen3-30B-A3B 71.9% May 2025 Lab model card
20 Qwen3-14B 71.7% May 2025 Lab model card
21 Qwen3.5-4B 71% Mar 2026 Lab model card
22 Qwen3-8B 67.8% May 2025 Lab model card
23 Granite 4.1 30B 67.3% Apr 2026 Lab model card
24 Granite 4.0 H Small 64% Oct 2025 Lab model card
25 Qwen3-4B 61.8% May 2025 Lab model card
26 Aya Expanse 32B 59.5% English instructions Oct 2024 Official leaderboard
27 Apertus 70B 58.2% Sep 2025 Lab model card
28 Gemma 3n E4B 57.2% Jun 2025 Lab model card
29 Qwen3.5-2B 55.4% Mar 2026 Lab model card
30 Qwen3-1.7B 51.8% May 2025 Lab model card
31 Aya Expanse 8B 47.8% English instructions Oct 2024 Official leaderboard
32 Mistral-7B-Instruct-v0.1 44.3% English instructions Sep 2023 Official leaderboard
33 granite-4.0-1b 42.1% Oct 2025 Lab model card
34 Qwen3.5-0.8B 40.6% Mar 2026 Lab model card

The benchmark's own leaderboard has 12 entries, 3 of them models we track. Switch the scores above to rank them together.

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Aya Expanse 32B
original-la…
59.5%
Aya Expanse 8B
original-la…
47.8%
Mistral-7B-Instruct-v0.1
original-la…
44.3%

Source: Official INCLUDE leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Aya Expanse 32B English instructions 59.5%
Aya Expanse 32B original-language instructions 59.1%
Aya Expanse 8B English instructions 47.8%
Aya Expanse 8B original-language instructions 47.2%
Mistral-7B-Instruct-v0.1 English instructions 44.3%
Mistral-7B-Instruct-v0.1 original-language instructions 44.2%

7 models have held number one in labs’ own numbers since its release in November 2024 Number one over time 8 models have held number one with claims included since its release in November 2024

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

Cohere Qwen Anthropic
Own leaderboard Not tested since Nov 2024
Aya Expanse 32B 59.5%
Lab claims Passed 85% in Feb 2026, by Qwen3.5 397B-A17B
Claude Opus 5 89.8%

Sources: Official INCLUDE leaderboard and labs’ model cards

Qwen’s own number for Qwen3.5 397B-A17B passed 85% in February 2026

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

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Claude Opus 5 Qwen3.5 397B-A17B Qwen3-235B-A22B-Thinking Cohere Command A Aya Expanse 32B Mistral-7B-Instruct-v0.1 0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Claude Opus 5 Qwen3-235B-A22B-Thinking

Sources: Official INCLUDE leaderboard and labs’ model cards

Labs headline it less often now, last in June 2026

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Qwen
Google DeepMind
IBM
DateLaunch postScoreAs the post writes it
Jun 1, 2026 Qwen3.7-Plus 83 INCLUDE
May 20, 2026 Qwen3.7-Max 86.2 INCLUDE
Apr 29, 2026 Granite 4.1 30B 67.3 INCLUDE (5-shot, 14 languages)
Apr 2, 2026 Qwen 3.6 Plus 85.1 INCLUDE
Mar 2, 2026 Qwen3.5-0.8B 40.6 INCLUDE
Mar 2, 2026 Qwen3.5-2B 55.4 INCLUDE
Mar 2, 2026 Qwen3.5-4B 71 INCLUDE
Mar 2, 2026 Qwen3.5-9B 75.6 INCLUDE
Feb 24, 2026 Qwen3.5-35B-A3B 79.7 INCLUDE
Feb 24, 2026 Qwen3.5-27B 81.6 INCLUDE
Feb 24, 2026 Qwen3.5-122B-A10B 82.8 INCLUDE
Feb 16, 2026 Qwen3.5 397B-A17B 85.6 INCLUDE
Oct 28, 2025 granite-4.0-1b 42.1 INCLUDE (5-shot, 14 languages)
Oct 2, 2025 Granite 4.0 H Small 64 INCLUDE (5-shot, 14 languages)
Sep 23, 2025 Qwen3-VL-235B-A22B 80 INCLUDE
Sep 11, 2025 Qwen3-Next-80B-A3B 78.9 INCLUDE
Jul 25, 2025 Qwen3-235B-A22B-Thinking 81 INCLUDE
Jun 26, 2025 Gemma 3n E4B 57.2 Include 0-shot (IT)
Apr 29, 2025 Qwen3-235B-A22B 78.7 INCLUDE

14 system cards report it, from 4 labs

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Jul 24, 2026 Claude Opus 5 system card 89.8 INCLUDE average accuracy across 44 languages, max effort, five trials, printed on chart (8.16.3, Figure 8.16.3.A) Capabilities
Jun 30, 2026 Claude Sonnet 5 system card 86.5 INCLUDE average accuracy across 44 languages, max effort, five trials (8.13.3, Figure 8.13.3.A) Capabilities
May 28, 2026 Claude Opus 4.8 system card 87.6 INCLUDE average accuracy across 44 languages, adaptive thinking (8.15.3, Figure 8.15.3.A) Capabilities
Apr 16, 2026 Claude Opus 4.7 system card 87 INCLUDE average accuracy across 44 languages, adaptive thinking (8.12.3, Figure 8.12.3.A) Capabilities
Nov 26, 2025 Qwen3-VL-235B-A22B technical report 80 INCLUDE (Table 6) Capabilities
Sep 17, 2025 Apertus 70B technical report 58.2 INCLUDE, 44 languages (Table 20). INCLUDE V2 beta, 45 more languages, 41.6 Capabilities
May 14, 2025 Qwen3-8B technical report 67.8 INCLUDE, thinking mode (Table 17). Non-thinking 62.5 (Table 18) Capabilities
May 14, 2025 Qwen3-4B technical report 61.8 INCLUDE, thinking mode (Table 17). Non-thinking 53.8 (Table 18) Capabilities
May 14, 2025 Qwen3-32B technical report 73.7 INCLUDE, thinking mode (Table 13). Non-thinking 70.9 (Table 14) Capabilities
May 14, 2025 Qwen3-30B-A3B technical report 71.9 INCLUDE, thinking mode (Table 15). Non-thinking 67.8 (Table 16) Capabilities
May 14, 2025 Qwen3-235B-A22B technical report 78.7 INCLUDE, thinking mode (Table 11). Non-thinking 75.6 (Table 12) Capabilities
May 14, 2025 Qwen3-14B technical report 71.7 INCLUDE, thinking mode (Table 15). Non-thinking 67.8 (Table 16) Capabilities
May 14, 2025 Qwen3-1.7B technical report 51.8 INCLUDE, thinking mode (Table 19). Non-thinking 42.6 (Table 20) Capabilities
Apr 1, 2025 Cohere Command A technical report 74.3 INCLUDE-44 average over 20 languages (Table 25) Capabilities

More in Multilingual

All 120 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. Nemotron 3 Ultra 550B-A55B Base 90.1%
  • 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%