Benchmark Almanac

Miðeind Icelandic ARC-Challenge

GPT-5.5 95.2% leads since April 2026
Not in model cards

Models answer multiple-choice science questions in an Icelandic translation of ARC-Challenge. Miðeind used Claude 3.5 Sonnet to translate the questions.

Passes test 2. Miðeind reports scores for 92 named models on its archived leaderboard. A human audit found translation errors and distinguished this version from the EuroEval translation.

Made by
Miðeind
Released
June 2024
Kind
Reasoning in many languages
Level
School
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
Icelandic
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Based on
ARC-Challenge
Strongest model tested
Claude Fable 5 #12 on the capability index
Last model added
August 2026
Scores refreshed
November 20th, 2026
90
Entries on its own leaderboard, 78 models we track

3 models share the top score of 95.2% among models released by August 2026 3 models share the top score of 95.2% among models released by August 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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 August 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

90 entries. Scroll sideways for the rest.

  1. 95.2 GPT-5.5
  2. 95.2 GPT-5.6 Sol
  3. 95.2 Gemini 3.7 Flash
  4. 95.1 Gemini 3.1 Pro
  5. 95.1 Gemini 3.5 Flash
  6. 95.0 Gemini 3 Pro
  7. 95.0 Gemini 3 Flash
  8. 95.0 Claude Opus 4.8
  9. 94.9 Gemini 3.6 Flash
  10. 94.8 Claude Opus 4.6
  11. 94.8 GPT-5.6 Terra
  12. 94.7 Gemma 4 31B IT
  13. 94.6 Gemini 3.1 Flash-Lite
  14. 94.6 GPT-5.4
  15. 94.6 Grok 4.3
  16. 94.5 GPT-5.1
  17. 94.5 Claude Opus 4.5
  18. 94.5 GLM-5.2
  19. 94.4 GPT-5
  20. 94.1 GPT-5.2
  21. 94.0 DeepSeek-V4-Pro
  22. 93.9 Gemini 2.5 Pro
  23. 93.4 o1-preview
  24. 93.4 Gemini 2.5 Flash
  25. 93.3 Claude Opus 4.7
  26. 93.1 Claude Sonnet 4.5
  27. 93.1 Gemma 4 26B A4B
  28. 93.0 o4-mini
  29. 93.0 GPT-5 mini
  30. 92.7 Claude Sonnet 4.6
  31. 92.6 Gemini 2.0 Pro
  32. 92.5 Claude Sonnet 4
  33. 92.4 GPT-5.6 Luna
  34. 92.1 Kimi K2.5
  35. 92.0 cyboghostginx/gemma-4-31B-it-Adetayo
  36. 92.0 Qwen3.5 397B-A17B
  37. 91.5 GPT-4.1
  38. 91.3 Claude 3.5 Sonnet
  39. 91.2 MiniMax-M2.5
  40. 90.8 Gemini 2.0 Flash
  41. 90.6 DeepSeek-R1
  42. 90.4 GPT-4o
  43. 90.4 Claude 3.7 Sonnet
  44. 90.4 GPT-5 nano
  45. 89.8 GPT-4o
  46. 89.2 o3-mini
  47. 88.7 GPT-4 Turbo
  48. 87.7 Gemini 2.5 Flash-Lite
  49. 87.2 GPT-4.1 mini
  50. 87.1 Mistral Large 3
  51. 86.1 o1-mini
  52. 85.2 Claude 3 Opus
  53. 84.2 cyboghostginx/gemma-3-27b-it-Adetayo
  54. 83.7 GPT-4
  55. 82.6 GPT-4
  56. 81.5 GPT-4o mini
  57. 80.7 Gemma 3 27B
  58. 79.7 DeepSeek-V3
  59. 78.4 DeepSeek-R1-Distill-Llama-70B
  60. 78.0 Claude 3.5 Haiku
  61. 77.9 Mistral Medium 3.1
  62. 75.3 Gemma 3 12B
  63. 74.7 Llama 3.3 70B Instruct
  64. 74.6 Llama-3.1-Nemotron-70B-Instruct
  65. 74.5 Llama 3.1-70B
  66. 73.0 Claude Fable 5
  67. 72.0 NousResearch/Hermes-3-Llama-3.1-405B-FP8
  68. 70.4 GPT-4.1 nano
  69. 60.8 Llama 3-70B
  70. 59.9 Claude 2.1
  71. 56.0 Qwen2-72B
  72. 49.7 cyboghostginx/gemma-3-4b-it-Adetayo
  73. 49.5 GPT-3.5 Turbo
  74. 48.8 Gemma 3 4B
  75. 41.9 Mistral NeMo
  76. 33.9 Llama 3.1-8B
  77. 29.4 Llama 3.2 3B
  78. 28.5 Ministral 3B
  79. 27.5 Llama 3-8B
  80. 26.3 Llama 3.2 1B
  81. 26.1 Gemma 3 1B
  82. 22.5 AI-Sweden-Models/gpt-sw3-20b-instruct
  83. 22.5 LumiOpen/Viking-7B
  84. 22.4 AI-Sweden-Models/gpt-sw3-1.3b-instruct
  85. 22.4 AI-Sweden-Models/gpt-sw3-6.7b-v2-instruct
  86. 22.4 LumiOpen/Viking-13B
  87. 21.8 LumiOpen/Viking-33B
  88. 16.0 Qwen2-7B-Instruct
  89. 15.0 AI-Sweden-Models/gpt-sw3-356m-instruct
  90. 7.51 google/gemma-2-9b-it
  1. 1 GPT-5.5 chat, high 95.2%
  2. 2 GPT-5.6 Sol chat, high 95.2%
  3. 3 Gemini 3.7 Flash chat, high 95.2%
  4. 4 Gemini 3.1 Pro chat, high 95.1%
  5. 5 Gemini 3.5 Flash chat, high 95.1%
  6. 6 Gemini 3 Pro chat, high 95%
  7. 7 Gemini 3 Flash chat, high 95%
  8. 8 Claude Opus 4.8 chat, high 95%
  9. 9 Gemini 3.6 Flash chat, high 94.9%
  10. 10 Claude Opus 4.6 chat, high 94.8%
  11. 11 GPT-5.6 Terra chat, high 94.8%
  12. 12 Gemma 4 31B IT chat, on 94.7%
  13. 13 Gemini 3.1 Flash-Lite chat, high 94.6%
  14. 14 GPT-5.4 chat, high 94.6%
  15. 15 Grok 4.3 chat, on 94.6%
  16. 16 GPT-5.1 chat, high 94.5%
  17. 17 Claude Opus 4.5 chat, high 94.5%
  18. 18 GLM-5.2 chat, on 94.5%
  19. 19 GPT-5 chat, high 94.4%
  20. 20 GPT-5.2 chat, high 94.1%
  21. 21 DeepSeek-V4-Pro chat, on 94%
  22. 22 Gemini 2.5 Pro Jun 2025, chat, on 93.9%
  23. 23 o1-preview chat, on 93.4%
  24. 24 Gemini 2.5 Flash Sep 2025, chat, high 93.4%
  25. 25 Claude Opus 4.7 chat, high 93.3%
  26. 26 Claude Sonnet 4.5 chat, high 93.1%
  27. 27 Gemma 4 26B A4B chat, on 93.1%
  28. 28 o4-mini chat, on 93%
  29. 29 GPT-5 mini chat, high 93%
  30. 30 Claude Sonnet 4.6 chat, high 92.7%
  31. 31 Gemini 2.0 Pro chat, on 92.6%
  32. 32 Claude Sonnet 4 chat, on 92.5%
  33. 33 GPT-5.6 Luna chat, high 92.4%
  34. 34 Kimi K2.5 chat, on 92.1%
  35. 35 cyboghostginx/gemma-4-31B-it-Adetayo fine-tuned 92%
  36. 36 Qwen3.5 397B-A17B chat, on 92%
  37. 37 GPT-4.1 chat 91.5%
  38. 38 Claude 3.5 Sonnet chat 91.3%
  39. 39 MiniMax-M2.5 chat, on 91.2%
  40. 40 Gemini 2.0 Flash Feb 2025, chat 90.8%
  41. 41 DeepSeek-R1 May 2025, chat, on 90.6%
  42. 42 GPT-4o Aug 2024, chat 90.4%
  43. 43 Claude 3.7 Sonnet chat 90.4%
  44. 44 GPT-5 nano chat, high 90.4%
  45. 45 GPT-4o May 2024, chat 89.8%
  46. 46 o3-mini chat, on 89.2%
  47. 47 GPT-4 Turbo Apr 2024, chat 88.7%
  48. 48 Gemini 2.5 Flash-Lite Sep 2025, chat, high 87.7%
  49. 49 GPT-4.1 mini chat 87.2%
  50. 50 Mistral Large 3 chat 87.1%
  51. 51 o1-mini chat, on 86.1%
  52. 52 Claude 3 Opus chat 85.2%
  53. 53 cyboghostginx/gemma-3-27b-it-Adetayo fine-tuned 84.2%
  54. 54 GPT-4 Jun 2023, chat 83.7%
  55. 55 GPT-4 Mar 2023, chat 82.6%
  56. 56 GPT-4o mini chat 81.5%
  57. 57 Gemma 3 27B chat 80.7%
  58. 58 DeepSeek-V3 Mar 2025, chat 79.7%
  59. 59 DeepSeek-R1-Distill-Llama-70B chat, on 78.4%
  60. 60 Claude 3.5 Haiku chat 78%
  61. 61 Mistral Medium 3.1 chat 77.9%
  62. 62 Gemma 3 12B chat 75.3%
  63. 63 Llama 3.3 70B Instruct chat 74.7%
  64. 64 Llama-3.1-Nemotron-70B-Instruct chat 74.6%
  65. 65 Llama 3.1-70B chat 74.5%
  66. 66 Claude Fable 5 chat, high 73%
  67. 67 NousResearch/Hermes-3-Llama-3.1-405B-FP8 chat 72%
  68. 68 GPT-4.1 nano chat 70.4%
  69. 69 Llama 3-70B chat 60.8%
  70. 70 Claude 2.1 chat 59.9%
  71. 71 Qwen2-72B chat 56%
  72. 72 cyboghostginx/gemma-3-4b-it-Adetayo fine-tuned 49.7%
  73. 73 GPT-3.5 Turbo Jan 2024, chat 49.5%
  74. 74 Gemma 3 4B chat 48.8%
  75. 75 Mistral NeMo chat 41.9%
  76. 76 Llama 3.1-8B chat 33.9%
  77. 77 Llama 3.2 3B chat 29.4%
  78. 78 Ministral 3B chat 28.5%
  79. 79 Llama 3-8B chat 27.5%
  80. 80 Llama 3.2 1B chat 26.3%
  81. 81 Gemma 3 1B chat 26.1%
  82. 82 AI-Sweden-Models/gpt-sw3-20b-instruct chat 22.5%
  83. 83 LumiOpen/Viking-7B pretrained 22.5%
  84. 84 AI-Sweden-Models/gpt-sw3-1.3b-instruct chat 22.4%
  85. 85 AI-Sweden-Models/gpt-sw3-6.7b-v2-instruct chat 22.4%
  86. 86 LumiOpen/Viking-13B pretrained 22.4%
  87. 87 LumiOpen/Viking-33B pretrained 21.8%
  88. 88 Qwen2-7B-Instruct chat 16%
  89. 89 AI-Sweden-Models/gpt-sw3-356m-instruct chat 15%
  90. 90 google/gemma-2-9b-it chat 7.51%

Source: Official Miðeind Icelandic ARC-Challenge leaderboard

  1. 1 GPT-5.5 chat, high 95.2%
  2. 2 GPT-5.6 Sol chat, high 95.2%
  3. 3 Gemini 3.7 Flash chat, high 95.2%
  4. 4 Gemini 3.1 Pro chat, high 95.1%
  5. 5 Gemini 3.5 Flash chat, high 95.1%
  6. 6 Gemini 3 Pro chat, high 95%
  7. 7 Gemini 3 Flash chat, high 95%
  8. 8 Claude Opus 4.8 chat, high 95%
  9. 9 Gemini 3.6 Flash chat, high 94.9%
  10. 10 Claude Opus 4.6 chat, high 94.8%
  11. 11 GPT-5.6 Terra chat, high 94.8%
  12. 12 Gemma 4 31B IT chat, on 94.7%
  13. 13 Gemini 3.1 Flash-Lite chat, high 94.6%
  14. 14 GPT-5.4 chat, high 94.6%
  15. 15 Grok 4.3 chat, on 94.6%
  16. 16 GPT-5.1 chat, high 94.5%
  17. 17 Claude Opus 4.5 chat, high 94.5%
  18. 18 GLM-5.2 chat, on 94.5%
  19. 19 GPT-5 chat, high 94.4%
  20. 20 GPT-5.2 chat, high 94.1%
  21. 21 DeepSeek-V4-Pro chat, on 94%
  22. 22 Gemini 2.5 Pro Jun 2025, chat, on 93.9%
  23. 23 o1-preview chat, on 93.4%
  24. 24 Gemini 2.5 Flash Sep 2025, chat, high 93.4%
  25. 25 Claude Opus 4.7 chat, high 93.3%
  26. 26 Claude Sonnet 4.5 chat, high 93.1%
  27. 27 Gemma 4 26B A4B chat, on 93.1%
  28. 28 o4-mini chat, on 93%
  29. 29 GPT-5 mini chat, high 93%
  30. 30 Claude Sonnet 4.6 chat, high 92.7%
  31. 31 Gemini 2.0 Pro chat, on 92.6%
  32. 32 Claude Sonnet 4 chat, on 92.5%
  33. 33 GPT-5.6 Luna chat, high 92.4%
  34. 34 Kimi K2.5 chat, on 92.1%
  35. 35 Qwen3.5 397B-A17B chat, on 92%
  36. 36 GPT-4.1 chat 91.5%
  37. 37 Claude 3.5 Sonnet chat 91.3%
  38. 38 MiniMax-M2.5 chat, on 91.2%
  39. 39 Gemini 2.0 Flash Feb 2025, chat 90.8%
  40. 40 DeepSeek-R1 May 2025, chat, on 90.6%
  41. 41 GPT-4o Aug 2024, chat 90.4%
  42. 42 Claude 3.7 Sonnet chat 90.4%
  43. 43 GPT-5 nano chat, high 90.4%
  44. 44 GPT-4o May 2024, chat 89.8%
  45. 45 o3-mini chat, on 89.2%
  46. 46 GPT-4 Turbo Apr 2024, chat 88.7%
  47. 47 Gemini 2.5 Flash-Lite Sep 2025, chat, high 87.7%
  48. 48 GPT-4.1 mini chat 87.2%
  49. 49 Mistral Large 3 chat 87.1%
  50. 50 o1-mini chat, on 86.1%
  51. 51 Claude 3 Opus chat 85.2%
  52. 52 GPT-4 Jun 2023, chat 83.7%
  53. 53 GPT-4 Mar 2023, chat 82.6%
  54. 54 GPT-4o mini chat 81.5%
  55. 55 Gemma 3 27B chat 80.7%
  56. 56 DeepSeek-V3 Mar 2025, chat 79.7%
  57. 57 DeepSeek-R1-Distill-Llama-70B chat, on 78.4%
  58. 58 Claude 3.5 Haiku chat 78%
  59. 59 Mistral Medium 3.1 chat 77.9%
  60. 60 Gemma 3 12B chat 75.3%
  61. 61 Llama 3.3 70B Instruct chat 74.7%
  62. 62 Llama-3.1-Nemotron-70B-Instruct chat 74.6%
  63. 63 Llama 3.1-70B chat 74.5%
  64. 64 Claude Fable 5 chat, high 73%
  65. 65 GPT-4.1 nano chat 70.4%
  66. 66 Llama 3-70B chat 60.8%
  67. 67 Claude 2.1 chat 59.9%
  68. 68 Qwen2-72B chat 56%
  69. 69 GPT-3.5 Turbo Jan 2024, chat 49.5%
  70. 70 Gemma 3 4B chat 48.8%
  71. 71 Mistral NeMo chat 41.9%
  72. 72 Llama 3.1-8B chat 33.9%
  73. 73 Llama 3.2 3B chat 29.4%
  74. 74 Ministral 3B chat 28.5%
  75. 75 Llama 3-8B chat 27.5%
  76. 76 Llama 3.2 1B chat 26.3%
  77. 77 Gemma 3 1B chat 26.1%
  78. 78 Qwen2-7B-Instruct chat 16%

Source: Official Miðeind Icelandic ARC-Challenge leaderboard

#EntryScoreSettingDateLinkSource
1 GPT-5.5 95.2% chat, high Apr 2026 developers.openai.com Official leaderboard
2 GPT-5.6 Sol 95.2% chat, high Jul 2026 developers.openai.com Official leaderboard
3 Gemini 3.7 Flash 95.2% chat, high Aug 2026 ai.google.dev Official leaderboard
4 Gemini 3.1 Pro 95.1% chat, high Feb 2026 ai.google.dev Official leaderboard
5 Gemini 3.5 Flash 95.1% chat, high May 2026 ai.google.dev Official leaderboard
6 Gemini 3 Pro 95% chat, high Nov 2025 ai.google.dev Official leaderboard
7 Gemini 3 Flash 95% chat, high Dec 2025 ai.google.dev Official leaderboard
8 Claude Opus 4.8 95% chat, high May 2026 anthropic.com Official leaderboard
9 Gemini 3.6 Flash 94.9% chat, high Jul 2026 ai.google.dev Official leaderboard
10 Claude Opus 4.6 94.8% chat, high Feb 2026 anthropic.com Official leaderboard
11 GPT-5.6 Terra 94.8% chat, high Jul 2026 developers.openai.com Official leaderboard
12 Gemma 4 31B IT 94.7% chat, on Apr 2026 huggingface.co Official leaderboard
13 Gemini 3.1 Flash-Lite 94.6% chat, high Mar 2026 ai.google.dev Official leaderboard
14 GPT-5.4 94.6% chat, high Mar 2026 huggingface.co Official leaderboard
15 Grok 4.3 94.6% chat, on Apr 2026 docs.x.ai Official leaderboard
16 GPT-5.1 94.5% chat, high Nov 2025 huggingface.co Official leaderboard
17 Claude Opus 4.5 94.5% chat, high Nov 2025 huggingface.co Official leaderboard
18 GLM-5.2 94.5% chat, on Jun 2026 huggingface.co Official leaderboard
19 GPT-5 94.4% chat, high Aug 2025 huggingface.co Official leaderboard
20 GPT-5.2 94.1% chat, high Dec 2025 huggingface.co Official leaderboard
21 DeepSeek-V4-Pro 94% chat, on Apr 2026 huggingface.co Official leaderboard
22 Gemini 2.5 Pro Jun 2025 93.9% chat, on Jun 2025 huggingface.co Official leaderboard
23 o1-preview 93.4% chat, on Sep 2024 huggingface.co Official leaderboard
24 Gemini 2.5 Flash Sep 2025 93.4% chat, high Sep 2025 huggingface.co Official leaderboard
25 Claude Opus 4.7 93.3% chat, high Apr 2026 anthropic.com Official leaderboard
26 Claude Sonnet 4.5 93.1% chat, high Sep 2025 huggingface.co Official leaderboard
27 Gemma 4 26B A4B 93.1% chat, on Apr 2026 huggingface.co Official leaderboard
28 o4-mini 93% chat, on Apr 2025 huggingface.co Official leaderboard
29 GPT-5 mini 93% chat, high Aug 2025 huggingface.co Official leaderboard
30 Claude Sonnet 4.6 92.7% chat, high Feb 2026 anthropic.com Official leaderboard
31 Gemini 2.0 Pro 92.6% chat, on Feb 2025 huggingface.co Official leaderboard
32 Claude Sonnet 4 92.5% chat, on May 2025 huggingface.co Official leaderboard
33 GPT-5.6 Luna 92.4% chat, high Jul 2026 developers.openai.com Official leaderboard
34 Kimi K2.5 92.1% chat, on Jan 2026 kimi.com Official leaderboard
35 cyboghostginx/gemma-4-31B-it-Adetayo 31B 92% fine-tuned huggingface.co Official leaderboard
36 Qwen3.5 397B-A17B 92% chat, on Feb 2026 huggingface.co Official leaderboard
37 GPT-4.1 91.5% chat Apr 2025 huggingface.co Official leaderboard
38 Claude 3.5 Sonnet 91.3% chat Jun 2024 huggingface.co Official leaderboard
39 MiniMax-M2.5 91.2% chat, on Feb 2026 minimax.io Official leaderboard
40 Gemini 2.0 Flash Feb 2025 90.8% chat Feb 2025 huggingface.co Official leaderboard
41 DeepSeek-R1 May 2025 90.6% chat, on May 2025 huggingface.co Official leaderboard
42 GPT-4o Aug 2024 90.4% chat Aug 2024 huggingface.co Official leaderboard
43 Claude 3.7 Sonnet 90.4% chat Feb 2025 huggingface.co Official leaderboard
44 GPT-5 nano 90.4% chat, high Aug 2025 huggingface.co Official leaderboard
45 GPT-4o May 2024 89.8% chat May 2024 huggingface.co Official leaderboard
46 o3-mini 89.2% chat, on Jan 2025 huggingface.co Official leaderboard
47 GPT-4 Turbo Apr 2024 88.7% chat Apr 2024 huggingface.co Official leaderboard
48 Gemini 2.5 Flash-Lite Sep 2025 87.7% chat, high Sep 2025 huggingface.co Official leaderboard
49 GPT-4.1 mini 87.2% chat Apr 2025 huggingface.co Official leaderboard
50 Mistral Large 3 87.1% chat Dec 2025 huggingface.co Official leaderboard
51 o1-mini 86.1% chat, on Sep 2024 huggingface.co Official leaderboard
52 Claude 3 Opus 85.2% chat Feb 2024 huggingface.co Official leaderboard
53 cyboghostginx/gemma-3-27b-it-Adetayo 27B 84.2% fine-tuned huggingface.co Official leaderboard
54 GPT-4 Jun 2023 83.7% chat Jun 2023 huggingface.co Official leaderboard
55 GPT-4 Mar 2023 82.6% chat Mar 2023 huggingface.co Official leaderboard
56 GPT-4o mini 81.5% chat Jul 2024 huggingface.co Official leaderboard
57 Gemma 3 27B 80.7% chat Mar 2025 huggingface.co Official leaderboard
58 DeepSeek-V3 Mar 2025 79.7% chat Mar 2025 huggingface.co Official leaderboard
59 DeepSeek-R1-Distill-Llama-70B 78.4% chat, on Jan 2025 huggingface.co Official leaderboard
60 Claude 3.5 Haiku 78% chat Oct 2024 huggingface.co Official leaderboard
61 Mistral Medium 3.1 77.9% chat Aug 2025 huggingface.co Official leaderboard
62 Gemma 3 12B 75.3% chat Mar 2025 huggingface.co Official leaderboard
63 Llama 3.3 70B Instruct 74.7% chat Dec 2024 huggingface.co Official leaderboard
64 Llama-3.1-Nemotron-70B-Instruct 74.6% chat Oct 2024 huggingface.co Official leaderboard
65 Llama 3.1-70B 74.5% chat Jul 2024 huggingface.co Official leaderboard
66 Claude Fable 5 73% chat, high Jun 2026 anthropic.com Official leaderboard
67 NousResearch/Hermes-3-Llama-3.1-405B-FP8 405.9B 72% chat huggingface.co Official leaderboard
68 GPT-4.1 nano 70.4% chat Apr 2025 huggingface.co Official leaderboard
69 Llama 3-70B 60.8% chat Apr 2024 huggingface.co Official leaderboard
70 Claude 2.1 59.9% chat Nov 2023 huggingface.co Official leaderboard
71 Qwen2-72B 56% chat Jun 2024 huggingface.co Official leaderboard
72 cyboghostginx/gemma-3-4b-it-Adetayo 4B 49.7% fine-tuned huggingface.co Official leaderboard
73 GPT-3.5 Turbo Jan 2024 49.5% chat Jan 2024 huggingface.co Official leaderboard
74 Gemma 3 4B 48.8% chat Mar 2025 huggingface.co Official leaderboard
75 Mistral NeMo 41.9% chat Jul 2024 huggingface.co Official leaderboard
76 Llama 3.1-8B 33.9% chat Jul 2024 huggingface.co Official leaderboard
77 Llama 3.2 3B 29.4% chat Sep 2024 huggingface.co Official leaderboard
78 Ministral 3B 28.5% chat Oct 2024 huggingface.co Official leaderboard
79 Llama 3-8B 27.5% chat Apr 2024 huggingface.co Official leaderboard
80 Llama 3.2 1B 26.3% chat Sep 2024 huggingface.co Official leaderboard
81 Gemma 3 1B 26.1% chat Mar 2025 huggingface.co Official leaderboard
82 AI-Sweden-Models/gpt-sw3-20b-instruct 20.9B 22.5% chat huggingface.co Official leaderboard
83 LumiOpen/Viking-7B 7.6B 22.5% pretrained huggingface.co Official leaderboard
84 AI-Sweden-Models/gpt-sw3-1.3b-instruct 1.4B 22.4% chat huggingface.co Official leaderboard
85 AI-Sweden-Models/gpt-sw3-6.7b-v2-instruct 7.1B 22.4% chat huggingface.co Official leaderboard
86 LumiOpen/Viking-13B 14B 22.4% pretrained huggingface.co Official leaderboard
87 LumiOpen/Viking-33B 33.1B 21.8% pretrained huggingface.co Official leaderboard
88 Qwen2-7B-Instruct 16% chat Jun 2024 huggingface.co Official leaderboard
89 AI-Sweden-Models/gpt-sw3-356m-instruct 0.5B 15% chat huggingface.co Official leaderboard
90 google/gemma-2-9b-it 9.2B 7.51% chat huggingface.co Official leaderboard
#ModelScoreSettingDateLinkSource
1 GPT-5.5 95.2% chat, high Apr 2026 developers.openai.com Official leaderboard
2 GPT-5.6 Sol 95.2% chat, high Jul 2026 developers.openai.com Official leaderboard
3 Gemini 3.7 Flash 95.2% chat, high Aug 2026 ai.google.dev Official leaderboard
4 Gemini 3.1 Pro 95.1% chat, high Feb 2026 ai.google.dev Official leaderboard
5 Gemini 3.5 Flash 95.1% chat, high May 2026 ai.google.dev Official leaderboard
6 Gemini 3 Pro 95% chat, high Nov 2025 ai.google.dev Official leaderboard
7 Gemini 3 Flash 95% chat, high Dec 2025 ai.google.dev Official leaderboard
8 Claude Opus 4.8 95% chat, high May 2026 anthropic.com Official leaderboard
9 Gemini 3.6 Flash 94.9% chat, high Jul 2026 ai.google.dev Official leaderboard
10 Claude Opus 4.6 94.8% chat, high Feb 2026 anthropic.com Official leaderboard
11 GPT-5.6 Terra 94.8% chat, high Jul 2026 developers.openai.com Official leaderboard
12 Gemma 4 31B IT 94.7% chat, on Apr 2026 huggingface.co Official leaderboard
13 Gemini 3.1 Flash-Lite 94.6% chat, high Mar 2026 ai.google.dev Official leaderboard
14 GPT-5.4 94.6% chat, high Mar 2026 huggingface.co Official leaderboard
15 Grok 4.3 94.6% chat, on Apr 2026 docs.x.ai Official leaderboard
16 GPT-5.1 94.5% chat, high Nov 2025 huggingface.co Official leaderboard
17 Claude Opus 4.5 94.5% chat, high Nov 2025 huggingface.co Official leaderboard
18 GLM-5.2 94.5% chat, on Jun 2026 huggingface.co Official leaderboard
19 GPT-5 94.4% chat, high Aug 2025 huggingface.co Official leaderboard
20 GPT-5.2 94.1% chat, high Dec 2025 huggingface.co Official leaderboard
21 DeepSeek-V4-Pro 94% chat, on Apr 2026 huggingface.co Official leaderboard
22 Gemini 2.5 Pro Jun 2025 93.9% chat, on Jun 2025 huggingface.co Official leaderboard
23 o1-preview 93.4% chat, on Sep 2024 huggingface.co Official leaderboard
24 Gemini 2.5 Flash Sep 2025 93.4% chat, high Sep 2025 huggingface.co Official leaderboard
25 Claude Opus 4.7 93.3% chat, high Apr 2026 anthropic.com Official leaderboard
26 Claude Sonnet 4.5 93.1% chat, high Sep 2025 huggingface.co Official leaderboard
27 Gemma 4 26B A4B 93.1% chat, on Apr 2026 huggingface.co Official leaderboard
28 o4-mini 93% chat, on Apr 2025 huggingface.co Official leaderboard
29 GPT-5 mini 93% chat, high Aug 2025 huggingface.co Official leaderboard
30 Claude Sonnet 4.6 92.7% chat, high Feb 2026 anthropic.com Official leaderboard
31 Gemini 2.0 Pro 92.6% chat, on Feb 2025 huggingface.co Official leaderboard
32 Claude Sonnet 4 92.5% chat, on May 2025 huggingface.co Official leaderboard
33 GPT-5.6 Luna 92.4% chat, high Jul 2026 developers.openai.com Official leaderboard
34 Kimi K2.5 92.1% chat, on Jan 2026 kimi.com Official leaderboard
35 Qwen3.5 397B-A17B 92% chat, on Feb 2026 huggingface.co Official leaderboard
36 GPT-4.1 91.5% chat Apr 2025 huggingface.co Official leaderboard
37 Claude 3.5 Sonnet 91.3% chat Jun 2024 huggingface.co Official leaderboard
38 MiniMax-M2.5 91.2% chat, on Feb 2026 minimax.io Official leaderboard
39 Gemini 2.0 Flash Feb 2025 90.8% chat Feb 2025 huggingface.co Official leaderboard
40 DeepSeek-R1 May 2025 90.6% chat, on May 2025 huggingface.co Official leaderboard
41 GPT-4o Aug 2024 90.4% chat Aug 2024 huggingface.co Official leaderboard
42 Claude 3.7 Sonnet 90.4% chat Feb 2025 huggingface.co Official leaderboard
43 GPT-5 nano 90.4% chat, high Aug 2025 huggingface.co Official leaderboard
44 GPT-4o May 2024 89.8% chat May 2024 huggingface.co Official leaderboard
45 o3-mini 89.2% chat, on Jan 2025 huggingface.co Official leaderboard
46 GPT-4 Turbo Apr 2024 88.7% chat Apr 2024 huggingface.co Official leaderboard
47 Gemini 2.5 Flash-Lite Sep 2025 87.7% chat, high Sep 2025 huggingface.co Official leaderboard
48 GPT-4.1 mini 87.2% chat Apr 2025 huggingface.co Official leaderboard
49 Mistral Large 3 87.1% chat Dec 2025 huggingface.co Official leaderboard
50 o1-mini 86.1% chat, on Sep 2024 huggingface.co Official leaderboard
51 Claude 3 Opus 85.2% chat Feb 2024 huggingface.co Official leaderboard
52 GPT-4 Jun 2023 83.7% chat Jun 2023 huggingface.co Official leaderboard
53 GPT-4 Mar 2023 82.6% chat Mar 2023 huggingface.co Official leaderboard
54 GPT-4o mini 81.5% chat Jul 2024 huggingface.co Official leaderboard
55 Gemma 3 27B 80.7% chat Mar 2025 huggingface.co Official leaderboard
56 DeepSeek-V3 Mar 2025 79.7% chat Mar 2025 huggingface.co Official leaderboard
57 DeepSeek-R1-Distill-Llama-70B 78.4% chat, on Jan 2025 huggingface.co Official leaderboard
58 Claude 3.5 Haiku 78% chat Oct 2024 huggingface.co Official leaderboard
59 Mistral Medium 3.1 77.9% chat Aug 2025 huggingface.co Official leaderboard
60 Gemma 3 12B 75.3% chat Mar 2025 huggingface.co Official leaderboard
61 Llama 3.3 70B Instruct 74.7% chat Dec 2024 huggingface.co Official leaderboard
62 Llama-3.1-Nemotron-70B-Instruct 74.6% chat Oct 2024 huggingface.co Official leaderboard
63 Llama 3.1-70B 74.5% chat Jul 2024 huggingface.co Official leaderboard
64 Claude Fable 5 73% chat, high Jun 2026 anthropic.com Official leaderboard
65 GPT-4.1 nano 70.4% chat Apr 2025 huggingface.co Official leaderboard
66 Llama 3-70B 60.8% chat Apr 2024 huggingface.co Official leaderboard
67 Claude 2.1 59.9% chat Nov 2023 huggingface.co Official leaderboard
68 Qwen2-72B 56% chat Jun 2024 huggingface.co Official leaderboard
69 GPT-3.5 Turbo Jan 2024 49.5% chat Jan 2024 huggingface.co Official leaderboard
70 Gemma 3 4B 48.8% chat Mar 2025 huggingface.co Official leaderboard
71 Mistral NeMo 41.9% chat Jul 2024 huggingface.co Official leaderboard
72 Llama 3.1-8B 33.9% chat Jul 2024 huggingface.co Official leaderboard
73 Llama 3.2 3B 29.4% chat Sep 2024 huggingface.co Official leaderboard
74 Ministral 3B 28.5% chat Oct 2024 huggingface.co Official leaderboard
75 Llama 3-8B 27.5% chat Apr 2024 huggingface.co Official leaderboard
76 Llama 3.2 1B 26.3% chat Sep 2024 huggingface.co Official leaderboard
77 Gemma 3 1B 26.1% chat Mar 2025 huggingface.co Official leaderboard
78 Qwen2-7B-Instruct 16% chat Jun 2024 huggingface.co Official leaderboard

9 models have held number one on the official leaderboard since its release in June 2024

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

OpenAI Anthropic Google DeepMind
Own leaderboard Passed 85% in Jun 2024, by GPT-4o
GPT-5.5 95.2%

Source: Official Miðeind Icelandic ARC-Challenge leaderboard

Claude 3 Opus on the official leaderboard passed 85% in February 2024

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.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 Near the ceiling (85%) GPT-5.5 GPT-5 o1-preview GPT-4 Turbo GPT-4 0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 Near the ceiling (85%) GPT-5.5 GPT-4o

Source: Official Miðeind Icelandic ARC-Challenge 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

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (90 entries).
  • Pass Models scored 78 models we track, plus 12 other entries on its leaderboard.
  • Pass Newest score The newest is from August 2026, for Gemini 3.7 Flash.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Claude Fable 5 (#12).
  • Note Headroom Saturated. The best is 95.2%, past 85% since February 2024.
Leaderboard import 3 pass
  • Pass Importer We read the board on mideind-icelandic-llm-leaderboard.static.hf.space: 80 rows for models we track and 12 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