Benchmark Almanac

EKI Mälumäng

Gemini 3.1 Pro 50.8% leads since August 2026
Not in model cards

Models answer 800 Estonian quiz questions without answer choices. Politics and geography each contribute 80 questions. GPT-5.4 nano judges answers that miss an exact match.

The institute publishes accuracy for 93 models. The authors selected harder questions from two volumes by Lembit and Uno Ainsoo.

Released
August 2026
Kind
Knowledge in many languages
Level
Everyday
Scored
Graded by a model
Answers
Short answer
Measures
Capability
Language
Estonian
In and out
Text → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Run by
A host
Strongest model tested
Claude Fable 5.2 #2 on the capability index
Last model added
November 2026
Scores refreshed
November 20th, 2026
93
Entries on its own leaderboard, 89 models we track

Gemini 3.1 Pro leads with 50.8% Gemini 3.1 Pro leads with 50.8%

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.

93 entries. Scroll sideways for the rest.

  1. 50.8 Gemini 3.1 Pro
  2. 50.6 Gemini 4 Argon
  3. 50.3 Gemini 3.8 Flash
  4. 47.0 Gemini 3.7 Flash
  5. 45.8 GPT-6 Astra
  6. 43.9 GPT-5.6 Sol
  7. 42.3 Gemini 3.5 Flash
  8. 41.3 Gemini 3.6 Flash
  9. 39.1 Claude Fable 5.2
  10. 38.4 GPT-5.5
  11. 37.1 GPT-6 Sol
  12. 30.4 GPT-6.1 Terra
  13. 28.5 Claude Fable 5
  14. 27.1 Claude Fable 5.1
  15. 26.8 Gemini 3 Flash
  16. 26.0 Gemini 2.5 Pro
  17. 24.1 Muse Spark 1.3
  18. 24.1 Muse Spark Open
  19. 23.1 Muse Spark 1.2
  20. 22.3 GPT-6 Luna
  21. 21.3 Claude Opus 5.5
  22. 21.0 Claude Opus 5
  23. 20.1 GPT-5.6 Terra
  24. 19.3 GPT-5
  25. 18.6 GPT-5.6 Luna
  26. 18.5 Grok 4.5
  27. 17.6 Grok 4.7
  28. 16.9 Grok 4.6
  29. 12.5 Qwen3.7-Max
  30. 12.3 GPT-5.2
  31. 12.1 DeepSeek-V4-Pro
  32. 11.4 DeepSeek V4.1 Flash
  33. 10.9 Claude Sonnet 5.5
  34. 10.0 Kimi K3
  35. 8.62 Claude Opus 4.7
  36. 8.62 Claude Opus 4.8
  37. 7.25 GLM-5.3
  38. 7.12 Claude Opus 4.6
  39. 7.12 GLM-5
  40. 6.75 Qwen 3.8 Max
  41. 6.25 GLM-5.3-Flash
  42. 6.12 Claude Opus 4
  43. 6.12 GLM-5.2
  44. 5.88 Claude Opus 4.5
  45. 5.25 Tencent Hy4 preview
  46. 5.12 DeepSeek-V4-Flash
  47. 4.88 GPT-5.4
  48. 4.62 Mistral Large 4
  49. 4.50 GLM-4.7
  50. 4.38 GPT-4.1
  51. 4.25 GPT-4o
  52. 4.25 Claude Sonnet 5
  53. 3.50 Qwen 3.6 Plus
  54. 3.50 Claude Haiku 5.5
  55. 3.38 GPT-4
  56. 3.12 GPT-5.4 Mini
  57. 3.12 Nemotron 3 Ultra
  58. 3.00 MiMo-V2.6 Pro
  59. 2.88 MiniMax-M3
  60. 2.75 Llama 4 Maverick
  61. 2.75 Claude Sonnet 4.6
  62. 2.75 Qwen3.8 Flash
  63. 2.62 Llama 3.1 70B Instruct
  64. 2.38 Claude Sonnet 4
  65. 2.25 DeepSeek-V3.2
  66. 2.12 Muse Glimmer
  67. 2.00 Mistral Large 2512
  68. 2.00 gpt-oss-120b
  69. 2.00 Kimi K2
  70. 2.00 Gemini 2.5 Flash
  71. 1.88 Llama 3.3 70B Instruct
  72. 1.75 Nemotron 3 Super 120B
  73. 1.38 Gemma 2 27B
  74. 1.38 MiMo-V2.6 Flash
  75. 1.25 Gemma 4 31B IT
  76. 1.25 Step 3.7 Flash
  77. 1.12 Command R+ (08-2024)
  78. 1.12 GPT-3.5 Turbo
  79. 1.12 DeepSeek-V3
  80. 1.12 Claude Sonnet 4.5
  81. 1.12 Qwen3.5-27B
  82. 1.12 MiniMax-M2.7
  83. 1.12 Mistral Medium 3.5
  84. 1.00 Mistral Small 4
  85. 0.88 Gemma 3 27B
  86. 0.88 Step 3.5 Flash
  87. 0.88 Qwen3.6 27B
  88. 0.75 Claude 3 Haiku
  89. 0.75 Claude Haiku 4.5
  90. 0.75 Grok 4.20
  91. 0.62 Laguna S 2.1
  92. 0.62 Qwen 3.8 27B
  93. 0.50 GPT-4o mini
  1. 1 Gemini 3.1 Pro 50.8%
  2. 2 Gemini 4 Argon 50.6%
  3. 3 Gemini 3.8 Flash 50.3%
  4. 4 Gemini 3.7 Flash 47%
  5. 5 GPT-6 Astra 45.8%
  6. 6 GPT-5.6 Sol 43.9%
  7. 7 Gemini 3.5 Flash 42.3%
  8. 8 Gemini 3.6 Flash 41.3%
  9. 9 Claude Fable 5.2 39.1%
  10. 10 GPT-5.5 38.4%
  11. 11 GPT-6 Sol 37.1%
  12. 12 GPT-6.1 Terra 30.4%
  13. 13 Claude Fable 5 28.5%
  14. 14 Claude Fable 5.1 27.1%
  15. 15 Gemini 3 Flash 26.8%
  16. 16 Gemini 2.5 Pro Jun 2025 26%
  17. 17 Muse Spark 1.3 24.1%
  18. 18 Muse Spark Open 24.1%
  19. 19 Muse Spark 1.2 23.1%
  20. 20 GPT-6 Luna 22.3%
  21. 21 Claude Opus 5.5 21.3%
  22. 22 Claude Opus 5 21%
  23. 23 GPT-5.6 Terra 20.1%
  24. 24 GPT-5 19.3%
  25. 25 GPT-5.6 Luna 18.6%
  26. 26 Grok 4.5 18.5%
  27. 27 Grok 4.7 17.6%
  28. 28 Grok 4.6 16.9%
  29. 29 Qwen3.7-Max 12.5%
  30. 30 GPT-5.2 12.3%
  31. 31 DeepSeek-V4-Pro 12.1%
  32. 32 DeepSeek V4.1 Flash 11.4%
  33. 33 Claude Sonnet 5.5 10.9%
  34. 34 Kimi K3 10%
  35. 35 Claude Opus 4.7 8.62%
  36. 36 Claude Opus 4.8 8.62%
  37. 37 GLM-5.3 7.25%
  38. 38 Claude Opus 4.6 7.12%
  39. 39 GLM-5 7.12%
  40. 40 Qwen 3.8 Max 6.75%
  41. 41 GLM-5.3-Flash 6.25%
  42. 42 Claude Opus 4 6.12%
  43. 43 GLM-5.2 6.12%
  44. 44 Claude Opus 4.5 5.88%
  45. 45 Tencent Hy4 preview 5.25%
  46. 46 DeepSeek-V4-Flash 5.12%
  47. 47 GPT-5.4 4.88%
  48. 48 Mistral Large 4 4.62%
  49. 49 GLM-4.7 4.50%
  50. 50 GPT-4.1 4.38%
  51. 51 GPT-4o Mar 2025 4.25%
  52. 52 Claude Sonnet 5 4.25%
  53. 53 Qwen 3.6 Plus 3.50%
  54. 54 Claude Haiku 5.5 3.50%
  55. 55 GPT-4 Jun 2023 3.38%
  56. 56 GPT-5.4 Mini 3.12%
  57. 57 Nemotron 3 Ultra 3.12%
  58. 58 MiMo-V2.6 Pro 3.00%
  59. 59 MiniMax-M3 2.88%
  60. 60 Llama 4 Maverick 2.75%
  61. 61 Claude Sonnet 4.6 2.75%
  62. 62 Qwen3.8 Flash 2.75%
  63. 63 Llama 3.1 70B Instruct 2.62%
  64. 64 Claude Sonnet 4 2.38%
  65. 65 DeepSeek-V3.2 2.25%
  66. 66 Muse Glimmer 2.12%
  67. 67 Mistral Large 2512 2.00%
  68. 68 gpt-oss-120b 2.00%
  69. 69 Kimi K2 Sep 2025 2.00%
  70. 70 Gemini 2.5 Flash Sep 2025 2.00%
  71. 71 Llama 3.3 70B Instruct 1.88%
  72. 72 Nemotron 3 Super 120B 1.75%
  73. 73 Gemma 2 27B 1.38%
  74. 74 MiMo-V2.6 Flash 1.38%
  75. 75 Gemma 4 31B IT 1.25%
  76. 76 Step 3.7 Flash 1.25%
  77. 77 Command R+ (08-2024) 1.12%
  78. 78 GPT-3.5 Turbo Jan 2024 1.12%
  79. 79 DeepSeek-V3 Mar 2025 1.12%
  80. 80 Claude Sonnet 4.5 1.12%
  81. 81 Qwen3.5-27B 1.12%
  82. 82 MiniMax-M2.7 1.12%
  83. 83 Mistral Medium 3.5 1.12%
  84. 84 Mistral Small 4 1.00%
  85. 85 Gemma 3 27B 0.88%
  86. 86 Step 3.5 Flash 0.88%
  87. 87 Qwen3.6 27B 0.88%
  88. 88 Claude 3 Haiku 0.75%
  89. 89 Claude Haiku 4.5 0.75%
  90. 90 Grok 4.20 0.75%
  91. 91 Laguna S 2.1 0.62%
  92. 92 Qwen 3.8 27B 0.62%
  93. 93 GPT-4o mini 0.50%

Source: Official EKI Mälumäng leaderboard

89 models. Scroll sideways for the rest.

  1. 50.8 Gemini 3.1 Pro
  2. 50.6 Gemini 4 Argon
  3. 50.3 Gemini 3.8 Flash
  4. 47.0 Gemini 3.7 Flash
  5. 45.8 GPT-6 Astra
  6. 43.9 GPT-5.6 Sol
  7. 42.3 Gemini 3.5 Flash
  8. 41.3 Gemini 3.6 Flash
  9. 39.1 Claude Fable 5.2
  10. 38.4 GPT-5.5
  11. 37.1 GPT-6 Sol
  12. 30.4 GPT-6.1 Terra
  13. 28.5 Claude Fable 5
  14. 27.1 Claude Fable 5.1
  15. 26.8 Gemini 3 Flash
  16. 26.0 Gemini 2.5 Pro
  17. 24.1 Muse Spark 1.3
  18. 24.1 Muse Spark Open
  19. 23.1 Muse Spark 1.2
  20. 22.3 GPT-6 Luna
  21. 21.3 Claude Opus 5.5
  22. 21.0 Claude Opus 5
  23. 20.1 GPT-5.6 Terra
  24. 19.3 GPT-5
  25. 18.6 GPT-5.6 Luna
  26. 18.5 Grok 4.5
  27. 17.6 Grok 4.7
  28. 16.9 Grok 4.6
  29. 12.5 Qwen3.7-Max
  30. 12.3 GPT-5.2
  31. 12.1 DeepSeek-V4-Pro
  32. 11.4 DeepSeek V4.1 Flash
  33. 10.9 Claude Sonnet 5.5
  34. 10.0 Kimi K3
  35. 8.62 Claude Opus 4.7
  36. 8.62 Claude Opus 4.8
  37. 7.25 GLM-5.3
  38. 7.12 Claude Opus 4.6
  39. 7.12 GLM-5
  40. 6.75 Qwen 3.8 Max
  41. 6.25 GLM-5.3-Flash
  42. 6.12 Claude Opus 4
  43. 6.12 GLM-5.2
  44. 5.88 Claude Opus 4.5
  45. 5.25 Tencent Hy4 preview
  46. 5.12 DeepSeek-V4-Flash
  47. 4.88 GPT-5.4
  48. 4.62 Mistral Large 4
  49. 4.50 GLM-4.7
  50. 4.38 GPT-4.1
  51. 4.25 GPT-4o
  52. 4.25 Claude Sonnet 5
  53. 3.50 Qwen 3.6 Plus
  54. 3.50 Claude Haiku 5.5
  55. 3.38 GPT-4
  56. 3.12 GPT-5.4 Mini
  57. 3.12 Nemotron 3 Ultra
  58. 3.00 MiMo-V2.6 Pro
  59. 2.88 MiniMax-M3
  60. 2.75 Llama 4 Maverick
  61. 2.75 Claude Sonnet 4.6
  62. 2.75 Qwen3.8 Flash
  63. 2.38 Claude Sonnet 4
  64. 2.25 DeepSeek-V3.2
  65. 2.12 Muse Glimmer
  66. 2.00 gpt-oss-120b
  67. 2.00 Kimi K2
  68. 2.00 Gemini 2.5 Flash
  69. 1.88 Llama 3.3 70B Instruct
  70. 1.38 Gemma 2 27B
  71. 1.38 MiMo-V2.6 Flash
  72. 1.25 Gemma 4 31B IT
  73. 1.25 Step 3.7 Flash
  74. 1.12 GPT-3.5 Turbo
  75. 1.12 DeepSeek-V3
  76. 1.12 Claude Sonnet 4.5
  77. 1.12 Qwen3.5-27B
  78. 1.12 MiniMax-M2.7
  79. 1.12 Mistral Medium 3.5
  80. 1.00 Mistral Small 4
  81. 0.88 Gemma 3 27B
  82. 0.88 Step 3.5 Flash
  83. 0.88 Qwen3.6 27B
  84. 0.75 Claude 3 Haiku
  85. 0.75 Claude Haiku 4.5
  86. 0.75 Grok 4.20
  87. 0.62 Laguna S 2.1
  88. 0.62 Qwen 3.8 27B
  89. 0.50 GPT-4o mini
  1. 1 Gemini 3.1 Pro 50.8%
  2. 2 Gemini 4 Argon 50.6%
  3. 3 Gemini 3.8 Flash 50.3%
  4. 4 Gemini 3.7 Flash 47%
  5. 5 GPT-6 Astra 45.8%
  6. 6 GPT-5.6 Sol 43.9%
  7. 7 Gemini 3.5 Flash 42.3%
  8. 8 Gemini 3.6 Flash 41.3%
  9. 9 Claude Fable 5.2 39.1%
  10. 10 GPT-5.5 38.4%
  11. 11 GPT-6 Sol 37.1%
  12. 12 GPT-6.1 Terra 30.4%
  13. 13 Claude Fable 5 28.5%
  14. 14 Claude Fable 5.1 27.1%
  15. 15 Gemini 3 Flash 26.8%
  16. 16 Gemini 2.5 Pro Jun 2025 26%
  17. 17 Muse Spark 1.3 24.1%
  18. 18 Muse Spark Open 24.1%
  19. 19 Muse Spark 1.2 23.1%
  20. 20 GPT-6 Luna 22.3%
  21. 21 Claude Opus 5.5 21.3%
  22. 22 Claude Opus 5 21%
  23. 23 GPT-5.6 Terra 20.1%
  24. 24 GPT-5 19.3%
  25. 25 GPT-5.6 Luna 18.6%
  26. 26 Grok 4.5 18.5%
  27. 27 Grok 4.7 17.6%
  28. 28 Grok 4.6 16.9%
  29. 29 Qwen3.7-Max 12.5%
  30. 30 GPT-5.2 12.3%
  31. 31 DeepSeek-V4-Pro 12.1%
  32. 32 DeepSeek V4.1 Flash 11.4%
  33. 33 Claude Sonnet 5.5 10.9%
  34. 34 Kimi K3 10%
  35. 35 Claude Opus 4.7 8.62%
  36. 36 Claude Opus 4.8 8.62%
  37. 37 GLM-5.3 7.25%
  38. 38 Claude Opus 4.6 7.12%
  39. 39 GLM-5 7.12%
  40. 40 Qwen 3.8 Max 6.75%
  41. 41 GLM-5.3-Flash 6.25%
  42. 42 Claude Opus 4 6.12%
  43. 43 GLM-5.2 6.12%
  44. 44 Claude Opus 4.5 5.88%
  45. 45 Tencent Hy4 preview 5.25%
  46. 46 DeepSeek-V4-Flash 5.12%
  47. 47 GPT-5.4 4.88%
  48. 48 Mistral Large 4 4.62%
  49. 49 GLM-4.7 4.50%
  50. 50 GPT-4.1 4.38%
  51. 51 GPT-4o Mar 2025 4.25%
  52. 52 Claude Sonnet 5 4.25%
  53. 53 Qwen 3.6 Plus 3.50%
  54. 54 Claude Haiku 5.5 3.50%
  55. 55 GPT-4 Jun 2023 3.38%
  56. 56 GPT-5.4 Mini 3.12%
  57. 57 Nemotron 3 Ultra 3.12%
  58. 58 MiMo-V2.6 Pro 3.00%
  59. 59 MiniMax-M3 2.88%
  60. 60 Llama 4 Maverick 2.75%
  61. 61 Claude Sonnet 4.6 2.75%
  62. 62 Qwen3.8 Flash 2.75%
  63. 63 Claude Sonnet 4 2.38%
  64. 64 DeepSeek-V3.2 2.25%
  65. 65 Muse Glimmer 2.12%
  66. 66 gpt-oss-120b 2.00%
  67. 67 Kimi K2 Sep 2025 2.00%
  68. 68 Gemini 2.5 Flash Sep 2025 2.00%
  69. 69 Llama 3.3 70B Instruct 1.88%
  70. 70 Gemma 2 27B 1.38%
  71. 71 MiMo-V2.6 Flash 1.38%
  72. 72 Gemma 4 31B IT 1.25%
  73. 73 Step 3.7 Flash 1.25%
  74. 74 GPT-3.5 Turbo Jan 2024 1.12%
  75. 75 DeepSeek-V3 Mar 2025 1.12%
  76. 76 Claude Sonnet 4.5 1.12%
  77. 77 Qwen3.5-27B 1.12%
  78. 78 MiniMax-M2.7 1.12%
  79. 79 Mistral Medium 3.5 1.12%
  80. 80 Mistral Small 4 1.00%
  81. 81 Gemma 3 27B 0.88%
  82. 82 Step 3.5 Flash 0.88%
  83. 83 Qwen3.6 27B 0.88%
  84. 84 Claude 3 Haiku 0.75%
  85. 85 Claude Haiku 4.5 0.75%
  86. 86 Grok 4.20 0.75%
  87. 87 Laguna S 2.1 0.62%
  88. 88 Qwen 3.8 27B 0.62%
  89. 89 GPT-4o mini 0.50%

Source: Official EKI Mälumäng leaderboard

#EntryScoreSettingDateSource
1 Gemini 3.1 Pro 50.8% Feb 2026 Official leaderboard
2 Gemini 4 Argon 50.6% Sep 2026 Official leaderboard
3 Gemini 3.8 Flash 50.3% Sep 2026 Official leaderboard
4 Gemini 3.7 Flash 47% Aug 2026 Official leaderboard
5 GPT-6 Astra 45.8% Sep 2026 Official leaderboard
6 GPT-5.6 Sol 43.9% Jul 2026 Official leaderboard
7 Gemini 3.5 Flash 42.3% May 2026 Official leaderboard
8 Gemini 3.6 Flash 41.3% Jul 2026 Official leaderboard
9 Claude Fable 5.2 39.1% Nov 2026 Official leaderboard
10 GPT-5.5 38.4% Apr 2026 Official leaderboard
11 GPT-6 Sol 37.1% Sep 2026 Official leaderboard
12 GPT-6.1 Terra 30.4% Oct 2026 Official leaderboard
13 Claude Fable 5 28.5% Jun 2026 Official leaderboard
14 Claude Fable 5.1 27.1% Sep 2026 Official leaderboard
15 Gemini 3 Flash 26.8% Dec 2025 Official leaderboard
16 Gemini 2.5 Pro Jun 2025 26% Jun 2025 Official leaderboard
17 Muse Spark 1.3 24.1% Sep 2026 Official leaderboard
18 Muse Spark Open 24.1% Oct 2026 Official leaderboard
19 Muse Spark 1.2 23.1% Aug 2026 Official leaderboard
20 GPT-6 Luna 22.3% Sep 2026 Official leaderboard
21 Claude Opus 5.5 21.3% Sep 2026 Official leaderboard
22 Claude Opus 5 21% Jul 2026 Official leaderboard
23 GPT-5.6 Terra 20.1% Jul 2026 Official leaderboard
24 GPT-5 19.3% Aug 2025 Official leaderboard
25 GPT-5.6 Luna 18.6% Jul 2026 Official leaderboard
26 Grok 4.5 18.5% Jul 2026 Official leaderboard
27 Grok 4.7 17.6% Sep 2026 Official leaderboard
28 Grok 4.6 16.9% Aug 2026 Official leaderboard
29 Qwen3.7-Max 12.5% May 2026 Official leaderboard
30 GPT-5.2 12.3% Dec 2025 Official leaderboard
31 DeepSeek-V4-Pro 12.1% Apr 2026 Official leaderboard
32 DeepSeek V4.1 Flash 11.4% Sep 2026 Official leaderboard
33 Claude Sonnet 5.5 10.9% Sep 2026 Official leaderboard
34 Kimi K3 10% Jul 2026 Official leaderboard
35 Claude Opus 4.7 8.62% Apr 2026 Official leaderboard
36 Claude Opus 4.8 8.62% May 2026 Official leaderboard
37 GLM-5.3 7.25% Aug 2026 Official leaderboard
38 Claude Opus 4.6 7.12% Feb 2026 Official leaderboard
39 GLM-5 7.12% Feb 2026 Official leaderboard
40 Qwen 3.8 Max 6.75% Aug 2026 Official leaderboard
41 GLM-5.3-Flash 6.25% Aug 2026 Official leaderboard
42 Claude Opus 4 6.12% May 2025 Official leaderboard
43 GLM-5.2 6.12% Jun 2026 Official leaderboard
44 Claude Opus 4.5 5.88% Nov 2025 Official leaderboard
45 Tencent Hy4 preview 5.25% Aug 2026 Official leaderboard
46 DeepSeek-V4-Flash 5.12% Apr 2026 Official leaderboard
47 GPT-5.4 4.88% Mar 2026 Official leaderboard
48 Mistral Large 4 4.62% Oct 2026 Official leaderboard
49 GLM-4.7 4.50% Dec 2025 Official leaderboard
50 GPT-4.1 4.38% Apr 2025 Official leaderboard
51 GPT-4o Mar 2025 4.25% Mar 2025 Official leaderboard
52 Claude Sonnet 5 4.25% Jun 2026 Official leaderboard
53 Qwen 3.6 Plus 3.50% Mar 2026 Official leaderboard
54 Claude Haiku 5.5 3.50% Oct 2026 Official leaderboard
55 GPT-4 Jun 2023 3.38% Jun 2023 Official leaderboard
56 GPT-5.4 Mini 3.12% Mar 2026 Official leaderboard
57 Nemotron 3 Ultra 3.12% Jun 2026 Official leaderboard
58 MiMo-V2.6 Pro 3.00% Sep 2026 Official leaderboard
59 MiniMax-M3 2.88% Jun 2026 Official leaderboard
60 Llama 4 Maverick 2.75% Apr 2025 Official leaderboard
61 Claude Sonnet 4.6 2.75% Feb 2026 Official leaderboard
62 Qwen3.8 Flash 2.75% Aug 2026 Official leaderboard
63 Llama 3.1 70B Instruct open weights 2.62% Official leaderboard
64 Claude Sonnet 4 2.38% May 2025 Official leaderboard
65 DeepSeek-V3.2 2.25% Dec 2025 Official leaderboard
66 Muse Glimmer 2.12% Aug 2026 Official leaderboard
67 Mistral Large 2512 open weights 2.00% Official leaderboard
68 gpt-oss-120b 2.00% Aug 2025 Official leaderboard
69 Kimi K2 Sep 2025 2.00% Sep 2025 Official leaderboard
70 Gemini 2.5 Flash Sep 2025 2.00% Sep 2025 Official leaderboard
71 Llama 3.3 70B Instruct 1.88% Dec 2024 Official leaderboard
72 Nemotron 3 Super 120B open weights 1.75% Official leaderboard
73 Gemma 2 27B 1.38% Jun 2024 Official leaderboard
74 MiMo-V2.6 Flash 1.38% Sep 2026 Official leaderboard
75 Gemma 4 31B IT 1.25% Apr 2026 Official leaderboard
76 Step 3.7 Flash 1.25% May 2026 Official leaderboard
77 Command R+ (08-2024) 1.12% Official leaderboard
78 GPT-3.5 Turbo Jan 2024 1.12% Jan 2024 Official leaderboard
79 DeepSeek-V3 Mar 2025 1.12% Mar 2025 Official leaderboard
80 Claude Sonnet 4.5 1.12% Sep 2025 Official leaderboard
81 Qwen3.5-27B 1.12% Feb 2026 Official leaderboard
82 MiniMax-M2.7 1.12% Mar 2026 Official leaderboard
83 Mistral Medium 3.5 1.12% Apr 2026 Official leaderboard
84 Mistral Small 4 1.00% Mar 2026 Official leaderboard
85 Gemma 3 27B 0.88% Mar 2025 Official leaderboard
86 Step 3.5 Flash 0.88% Feb 2026 Official leaderboard
87 Qwen3.6 27B 0.88% Apr 2026 Official leaderboard
88 Claude 3 Haiku 0.75% Mar 2024 Official leaderboard
89 Claude Haiku 4.5 0.75% Oct 2025 Official leaderboard
90 Grok 4.20 0.75% Feb 2026 Official leaderboard
91 Laguna S 2.1 0.62% Jul 2026 Official leaderboard
92 Qwen 3.8 27B 0.62% Aug 2026 Official leaderboard
93 GPT-4o mini 0.50% Jul 2024 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 3.1 Pro 50.8% Feb 2026 Official leaderboard
2 Gemini 4 Argon 50.6% Sep 2026 Official leaderboard
3 Gemini 3.8 Flash 50.3% Sep 2026 Official leaderboard
4 Gemini 3.7 Flash 47% Aug 2026 Official leaderboard
5 GPT-6 Astra 45.8% Sep 2026 Official leaderboard
6 GPT-5.6 Sol 43.9% Jul 2026 Official leaderboard
7 Gemini 3.5 Flash 42.3% May 2026 Official leaderboard
8 Gemini 3.6 Flash 41.3% Jul 2026 Official leaderboard
9 Claude Fable 5.2 39.1% Nov 2026 Official leaderboard
10 GPT-5.5 38.4% Apr 2026 Official leaderboard
11 GPT-6 Sol 37.1% Sep 2026 Official leaderboard
12 GPT-6.1 Terra 30.4% Oct 2026 Official leaderboard
13 Claude Fable 5 28.5% Jun 2026 Official leaderboard
14 Claude Fable 5.1 27.1% Sep 2026 Official leaderboard
15 Gemini 3 Flash 26.8% Dec 2025 Official leaderboard
16 Gemini 2.5 Pro Jun 2025 26% Jun 2025 Official leaderboard
17 Muse Spark 1.3 24.1% Sep 2026 Official leaderboard
18 Muse Spark Open 24.1% Oct 2026 Official leaderboard
19 Muse Spark 1.2 23.1% Aug 2026 Official leaderboard
20 GPT-6 Luna 22.3% Sep 2026 Official leaderboard
21 Claude Opus 5.5 21.3% Sep 2026 Official leaderboard
22 Claude Opus 5 21% Jul 2026 Official leaderboard
23 GPT-5.6 Terra 20.1% Jul 2026 Official leaderboard
24 GPT-5 19.3% Aug 2025 Official leaderboard
25 GPT-5.6 Luna 18.6% Jul 2026 Official leaderboard
26 Grok 4.5 18.5% Jul 2026 Official leaderboard
27 Grok 4.7 17.6% Sep 2026 Official leaderboard
28 Grok 4.6 16.9% Aug 2026 Official leaderboard
29 Qwen3.7-Max 12.5% May 2026 Official leaderboard
30 GPT-5.2 12.3% Dec 2025 Official leaderboard
31 DeepSeek-V4-Pro 12.1% Apr 2026 Official leaderboard
32 DeepSeek V4.1 Flash 11.4% Sep 2026 Official leaderboard
33 Claude Sonnet 5.5 10.9% Sep 2026 Official leaderboard
34 Kimi K3 10% Jul 2026 Official leaderboard
35 Claude Opus 4.7 8.62% Apr 2026 Official leaderboard
36 Claude Opus 4.8 8.62% May 2026 Official leaderboard
37 GLM-5.3 7.25% Aug 2026 Official leaderboard
38 Claude Opus 4.6 7.12% Feb 2026 Official leaderboard
39 GLM-5 7.12% Feb 2026 Official leaderboard
40 Qwen 3.8 Max 6.75% Aug 2026 Official leaderboard
41 GLM-5.3-Flash 6.25% Aug 2026 Official leaderboard
42 Claude Opus 4 6.12% May 2025 Official leaderboard
43 GLM-5.2 6.12% Jun 2026 Official leaderboard
44 Claude Opus 4.5 5.88% Nov 2025 Official leaderboard
45 Tencent Hy4 preview 5.25% Aug 2026 Official leaderboard
46 DeepSeek-V4-Flash 5.12% Apr 2026 Official leaderboard
47 GPT-5.4 4.88% Mar 2026 Official leaderboard
48 Mistral Large 4 4.62% Oct 2026 Official leaderboard
49 GLM-4.7 4.50% Dec 2025 Official leaderboard
50 GPT-4.1 4.38% Apr 2025 Official leaderboard
51 GPT-4o Mar 2025 4.25% Mar 2025 Official leaderboard
52 Claude Sonnet 5 4.25% Jun 2026 Official leaderboard
53 Qwen 3.6 Plus 3.50% Mar 2026 Official leaderboard
54 Claude Haiku 5.5 3.50% Oct 2026 Official leaderboard
55 GPT-4 Jun 2023 3.38% Jun 2023 Official leaderboard
56 GPT-5.4 Mini 3.12% Mar 2026 Official leaderboard
57 Nemotron 3 Ultra 3.12% Jun 2026 Official leaderboard
58 MiMo-V2.6 Pro 3.00% Sep 2026 Official leaderboard
59 MiniMax-M3 2.88% Jun 2026 Official leaderboard
60 Llama 4 Maverick 2.75% Apr 2025 Official leaderboard
61 Claude Sonnet 4.6 2.75% Feb 2026 Official leaderboard
62 Qwen3.8 Flash 2.75% Aug 2026 Official leaderboard
63 Claude Sonnet 4 2.38% May 2025 Official leaderboard
64 DeepSeek-V3.2 2.25% Dec 2025 Official leaderboard
65 Muse Glimmer 2.12% Aug 2026 Official leaderboard
66 gpt-oss-120b 2.00% Aug 2025 Official leaderboard
67 Kimi K2 Sep 2025 2.00% Sep 2025 Official leaderboard
68 Gemini 2.5 Flash Sep 2025 2.00% Sep 2025 Official leaderboard
69 Llama 3.3 70B Instruct 1.88% Dec 2024 Official leaderboard
70 Gemma 2 27B 1.38% Jun 2024 Official leaderboard
71 MiMo-V2.6 Flash 1.38% Sep 2026 Official leaderboard
72 Gemma 4 31B IT 1.25% Apr 2026 Official leaderboard
73 Step 3.7 Flash 1.25% May 2026 Official leaderboard
74 GPT-3.5 Turbo Jan 2024 1.12% Jan 2024 Official leaderboard
75 DeepSeek-V3 Mar 2025 1.12% Mar 2025 Official leaderboard
76 Claude Sonnet 4.5 1.12% Sep 2025 Official leaderboard
77 Qwen3.5-27B 1.12% Feb 2026 Official leaderboard
78 MiniMax-M2.7 1.12% Mar 2026 Official leaderboard
79 Mistral Medium 3.5 1.12% Apr 2026 Official leaderboard
80 Mistral Small 4 1.00% Mar 2026 Official leaderboard
81 Gemma 3 27B 0.88% Mar 2025 Official leaderboard
82 Step 3.5 Flash 0.88% Feb 2026 Official leaderboard
83 Qwen3.6 27B 0.88% Apr 2026 Official leaderboard
84 Claude 3 Haiku 0.75% Mar 2024 Official leaderboard
85 Claude Haiku 4.5 0.75% Oct 2025 Official leaderboard
86 Grok 4.20 0.75% Feb 2026 Official leaderboard
87 Laguna S 2.1 0.62% Jul 2026 Official leaderboard
88 Qwen 3.8 27B 0.62% Aug 2026 Official leaderboard
89 GPT-4o mini 0.50% Jul 2024 Official leaderboard

Rankings by part

We read 10 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as ajalugu, poliitika and sport. Gemini 4 Argon leads 4 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

ajalugu

89 models we track have a score here.

  1. 1 Gemini 3.8 Flash ↑2 53.8%
  2. 2 Gemini 3.1 Pro ↓1 48.8%
  3. 3 Gemini 4 Argon ↓1 48.8%
  4. 4 Gemini 3.7 Flash 47.5%
  5. 5 GPT-6 Astra 47.5%
  6. 6 GPT-5.6 Sol 46.3%
  7. 7 Claude Fable 5.2 ↑2 42.5%
  8. 8 Gemini 3.6 Flash 41.3%
  9. 9 GPT-5.5 ↑1 40%
  10. 10 GPT-6 Sol ↑1 40%

Number one over time

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

Google DeepMind
Own leaderboard 1 leader

Source: Official EKI Mälumäng leaderboard

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
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 2027 Near the ceiling (85%) Gemini 3.1 Pro Gemini 3 Flash Claude Opus 4 GPT-4 0% 20% 40% 60% 80% 100% 2023 2024 2025 2026 2027 Near the ceiling (85%) Gemini 3.1 Pro Gemini 3 Flash

Source: Official EKI Mälumäng 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

10 of 11 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (93 entries).
  • Pass Models scored 89 models we track, plus 4 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Claude Fable 5.2.
  • Pass Frontier models 7 of the top 10 have a score, led by Claude Fable 5.2 (#2).
  • Pass Headroom The best is 50.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on xn--mdupuu-pxaa.eki.ee: 89 rows for models we track and 4 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