Benchmark Almanac

Miðeind Icelandic Generative Quality

GPT-6.2 Sol 87.7% leads since November 2026
Not in model cards

Models write paragraphs in Icelandic. Málfríður checks for grammar errors, and a dictionary lookup counts words absent from the Database of Modern Icelandic Inflection.

Passes test 2. Miðeind reports scores for 60 named models on its current leaderboard. The quality score combines grammatical and vocabulary measures.

Made by
Miðeind
Released
September 2026
Kind
Chat and instructions in many languages
Level
School
Scored
Checked automatically
Answers
Open-ended
Tasks
Kept private
Measures
Capability
Language
Icelandic
In and out
Text → Text
Interaction
Single turn
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
62
Entries on its own leaderboard, 59 models we track

GPT-6.2 Sol leads with 87.7% GPT-6.2 Sol leads with 87.7%

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. Every run is in the next chart. 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.

  1. 1 GPT-6.2 Sol high 87.7%
  2. 2 GPT-6.1 Terra off 84.3%
  3. 3 GPT-6 Sol high 83.8%
  4. 4 Claude Fable 5.2 low 82.9%
  5. 5 GPT-6 Luna high 82.3%
  6. 6 GPT-6 Astra high 82.2%
  7. 7 Gemini 4 Hydrogen high 79.4%
  8. 8 GPT-6.1 Sol low 78.8%
  9. 9 GPT-5.6 Luna off 78.7%
  10. 10 GPT-5.6 Sol off 78.7%
  11. 11 Gemini 4 Neon low 78.3%
  12. 12 Grok 4.7 low 77.8%
  13. 13 Gemini 4 Carbon low 75.6%
  14. 14 DeepSeek-V4-Pro on 75.6%
  15. 15 Claude Sonnet 5.5 high 74.9%
  16. 16 Gemini 4 Argon high 74.9%
  17. 17 GPT-5.6 Terra off 74.6%
  18. 18 Claude Opus 5.5 high 74.5%
  19. 19 Gemini 3.1 Pro high 74.3%
  20. 20 Claude Fable 5.1 high 73.7%
  21. 21 Claude Opus 5 off 72.3%
  22. 22 Gemini 3.7 Flash low 71.4%
  23. 23 Claude Opus 4.8 high 70.7%
  24. 24 Claude Sonnet 5 high 70.2%
  25. 25 Claude Haiku 5.5 high 69%
  26. 26 GPT-4o Mar 2025, none 68.6%
  27. 27 Gemini 3.8 Flash low 67.8%
  28. 28 DeepSeek V4.1 Flash on 65.6%
  29. 29 GPT-5.2 off 64.8%
  30. 30 Inkling 2 Small minimal 64%
  31. 31 Gemini 2.5 Pro Jun 2025, minimal 63.9%
  32. 32 Qwen3.8-2.4T-A95B on 62.8%
  33. 33 Gemini 3.5 Flash high 62.6%
  34. 34 Muse Spark Open off 61.7%
  35. 35 Mistral Large 4 high 61%
  36. 36 Bielik 11B v3.1 Instruct off 60.3%
  37. 37 Inkling high 57.9%
  38. 38 Kimi K3 on 57.6%
  39. 39 Gemini 3.5 Flash-Lite high 57.5%
  40. 40 Mistral Large 3 none 55.2%
  41. 41 Gemini 2.5 Flash-Lite Sep 2025, high 55%
  42. 42 MiniMax-M3 on 48.8%
  43. 43 GPT-4.1 nano none 48.7%
  44. 44 Mistral Small 4 off 45.6%
  45. 45 Inkling-Small high 43.4%
  46. 46 Apertus 1.5 70B on 42.6%
  47. 47 Gemma 4 26B A4B on 37.5%
  48. 48 Llama 3.1 70B Instruct none 37.3%
  49. 49 GPT-3.5 Turbo Jan 2024, none 36.1%
  50. 50 Claude Haiku 4.5 off 35.4%
  51. 51 Mistral Medium 3.5 high 34.4%
  52. 52 MiniMax-M3.1-Flash high 34%
  53. 53 Gemma 4 31B IT on 33.3%
  54. 54 Llama 3.1 8B Instruct none 31.4%
  55. 55 Qwen 3.8 27B off 30.9%
  56. 56 Apertus 8B Instruct none 29.1%
  57. 57 Gemma 4 12B on 25.9%
  58. 58 Apertus 1.5 8B on 25.3%
  59. 59 Gemma 4 E2B off 24.3%
  60. 60 Gemma 4 E4B off 24.1%
  61. 61 GLM-5.3-Flash on 21.7%
  62. 62 Ripple-65B Preview on 18.3%

Source: Official Miðeind Icelandic Generative Quality leaderboard

  1. 1 GPT-6.2 Sol high 87.7%
  2. 2 GPT-6.1 Terra off 84.3%
  3. 3 GPT-6 Sol high 83.8%
  4. 4 Claude Fable 5.2 low 82.9%
  5. 5 GPT-6 Luna high 82.3%
  6. 6 GPT-6 Astra high 82.2%
  7. 7 Gemini 4 Hydrogen high 79.4%
  8. 8 GPT-6.1 Sol low 78.8%
  9. 9 GPT-5.6 Luna off 78.7%
  10. 10 GPT-5.6 Sol off 78.7%
  11. 11 Gemini 4 Neon low 78.3%
  12. 12 Grok 4.7 low 77.8%
  13. 13 Gemini 4 Carbon low 75.6%
  14. 14 DeepSeek-V4-Pro on 75.6%
  15. 15 Claude Sonnet 5.5 high 74.9%
  16. 16 Gemini 4 Argon high 74.9%
  17. 17 GPT-5.6 Terra off 74.6%
  18. 18 Claude Opus 5.5 high 74.5%
  19. 19 Gemini 3.1 Pro high 74.3%
  20. 20 Claude Fable 5.1 high 73.7%
  21. 21 Claude Opus 5 off 72.3%
  22. 22 Gemini 3.7 Flash low 71.4%
  23. 23 Claude Opus 4.8 high 70.7%
  24. 24 Claude Sonnet 5 high 70.2%
  25. 25 Claude Haiku 5.5 high 69%
  26. 26 GPT-4o Mar 2025, none 68.6%
  27. 27 Gemini 3.8 Flash low 67.8%
  28. 28 DeepSeek V4.1 Flash on 65.6%
  29. 29 GPT-5.2 off 64.8%
  30. 30 Inkling 2 Small minimal 64%
  31. 31 Gemini 2.5 Pro Jun 2025, minimal 63.9%
  32. 32 Qwen3.8-2.4T-A95B on 62.8%
  33. 33 Gemini 3.5 Flash high 62.6%
  34. 34 Muse Spark Open off 61.7%
  35. 35 Mistral Large 4 high 61%
  36. 36 Inkling high 57.9%
  37. 37 Kimi K3 on 57.6%
  38. 38 Gemini 3.5 Flash-Lite high 57.5%
  39. 39 Mistral Large 3 none 55.2%
  40. 40 Gemini 2.5 Flash-Lite Sep 2025, high 55%
  41. 41 MiniMax-M3 on 48.8%
  42. 42 GPT-4.1 nano none 48.7%
  43. 43 Mistral Small 4 off 45.6%
  44. 44 Inkling-Small high 43.4%
  45. 45 Apertus 1.5 70B on 42.6%
  46. 46 Gemma 4 26B A4B on 37.5%
  47. 47 GPT-3.5 Turbo Jan 2024, none 36.1%
  48. 48 Claude Haiku 4.5 off 35.4%
  49. 49 Mistral Medium 3.5 high 34.4%
  50. 50 MiniMax-M3.1-Flash high 34%
  51. 51 Gemma 4 31B IT on 33.3%
  52. 52 Qwen 3.8 27B off 30.9%
  53. 53 Apertus 8B Instruct none 29.1%
  54. 54 Gemma 4 12B on 25.9%
  55. 55 Apertus 1.5 8B on 25.3%
  56. 56 Gemma 4 E2B off 24.3%
  57. 57 Gemma 4 E4B off 24.1%
  58. 58 GLM-5.3-Flash on 21.7%
  59. 59 Ripple-65B Preview on 18.3%

Source: Official Miðeind Icelandic Generative Quality leaderboard

#EntryScoreSettingReleasedRun onLinkSource
1 GPT-6.2 Sol 87.7% high Nov 2026 Nov 2026 developers.openai.com Official leaderboard
2 GPT-6.1 Terra 84.3% off Oct 2026 Oct 2026 developers.openai.com Official leaderboard
3 GPT-6 Sol 83.8% high Sep 2026 Sep 2026 developers.openai.com Official leaderboard
4 Claude Fable 5.2 82.9% low Nov 2026 Nov 2026 anthropic.com Official leaderboard
5 GPT-6 Luna 82.3% high Sep 2026 Sep 2026 developers.openai.com Official leaderboard
6 GPT-6 Astra 82.2% high Sep 2026 Sep 2026 developers.openai.com Official leaderboard
7 Gemini 4 Hydrogen 79.4% high Nov 2026 Nov 2026 ai.google.dev Official leaderboard
8 GPT-6.1 Sol 78.8% low Sep 2026 Sep 2026 developers.openai.com Official leaderboard
9 GPT-5.6 Luna 78.7% off Jul 2026 Sep 2026 developers.openai.com Official leaderboard
10 GPT-5.6 Sol 78.7% off Jul 2026 Sep 2026 developers.openai.com Official leaderboard
11 Gemini 4 Neon 78.3% low Nov 2026 Nov 2026 ai.google.dev Official leaderboard
12 Grok 4.7 77.8% low Sep 2026 Nov 2026 docs.x.ai Official leaderboard
13 Gemini 4 Carbon 75.6% low Oct 2026 Nov 2026 ai.google.dev Official leaderboard
14 DeepSeek-V4-Pro 75.6% on Apr 2026 Sep 2026 huggingface.co Official leaderboard
15 Claude Sonnet 5.5 74.9% high Sep 2026 Sep 2026 anthropic.com Official leaderboard
16 Gemini 4 Argon 74.9% high Sep 2026 Oct 2026 ai.google.dev Official leaderboard
17 GPT-5.6 Terra 74.6% off Jul 2026 Sep 2026 developers.openai.com Official leaderboard
18 Claude Opus 5.5 74.5% high Sep 2026 Sep 2026 anthropic.com Official leaderboard
19 Gemini 3.1 Pro 74.3% high Feb 2026 Sep 2026 ai.google.dev Official leaderboard
20 Claude Fable 5.1 73.7% high Sep 2026 Sep 2026 anthropic.com Official leaderboard
21 Claude Opus 5 72.3% off Jul 2026 Sep 2026 anthropic.com Official leaderboard
22 Gemini 3.7 Flash 71.4% low Aug 2026 Sep 2026 ai.google.dev Official leaderboard
23 Claude Opus 4.8 70.7% high May 2026 Sep 2026 anthropic.com Official leaderboard
24 Claude Sonnet 5 70.2% high Jun 2026 Sep 2026 anthropic.com Official leaderboard
25 Claude Haiku 5.5 69% high Oct 2026 Oct 2026 anthropic.com Official leaderboard
26 GPT-4o Mar 2025 68.6% none Mar 2025 Sep 2026 platform.openai.com Official leaderboard
27 Gemini 3.8 Flash 67.8% low Sep 2026 Sep 2026 ai.google.dev Official leaderboard
28 DeepSeek V4.1 Flash 65.6% on Sep 2026 Oct 2026 huggingface.co Official leaderboard
29 GPT-5.2 64.8% off Dec 2025 Sep 2026 platform.openai.com Official leaderboard
30 Inkling 2 Small 64% minimal Nov 2026 Nov 2026 huggingface.co Official leaderboard
31 Gemini 2.5 Pro Jun 2025 63.9% minimal Jun 2025 Sep 2026 ai.google.dev Official leaderboard
32 Qwen3.8-2.4T-A95B 62.8% on Aug 2026 Sep 2026 huggingface.co Official leaderboard
33 Gemini 3.5 Flash 62.6% high May 2026 Sep 2026 ai.google.dev Official leaderboard
34 Muse Spark Open 61.7% off Oct 2026 Oct 2026 huggingface.co Official leaderboard
35 Mistral Large 4 61% high Oct 2026 Oct 2026 mistral.ai Official leaderboard
36 Bielik 11B v3.1 Instruct 11.2B 60.3% off Nov 2026 huggingface.co Official leaderboard
37 Inkling 57.9% high Jul 2026 Sep 2026 huggingface.co Official leaderboard
38 Kimi K3 57.6% on Jul 2026 Sep 2026 huggingface.co Official leaderboard
39 Gemini 3.5 Flash-Lite 57.5% high Jul 2026 Sep 2026 ai.google.dev Official leaderboard
40 Mistral Large 3 55.2% none Dec 2025 Sep 2026 mistral.ai Official leaderboard
41 Gemini 2.5 Flash-Lite Sep 2025 55% high Sep 2025 Sep 2026 ai.google.dev Official leaderboard
42 MiniMax-M3 48.8% on Jun 2026 Sep 2026 huggingface.co Official leaderboard
43 GPT-4.1 nano 48.7% none Apr 2025 Sep 2026 platform.openai.com Official leaderboard
44 Mistral Small 4 45.6% off Mar 2026 Sep 2026 mistral.ai Official leaderboard
45 Inkling-Small 43.4% high Jul 2026 Sep 2026 huggingface.co Official leaderboard
46 Apertus 1.5 70B 42.6% on Jul 2026 Sep 2026 huggingface.co Official leaderboard
47 Gemma 4 26B A4B 37.5% on Apr 2026 Oct 2026 huggingface.co Official leaderboard
48 Llama 3.1 70B Instruct 70B 37.3% none Oct 2026 huggingface.co Official leaderboard
49 GPT-3.5 Turbo Jan 2024 36.1% none Jan 2024 Sep 2026 platform.openai.com Official leaderboard
50 Claude Haiku 4.5 35.4% off Oct 2025 Sep 2026 anthropic.com Official leaderboard
51 Mistral Medium 3.5 34.4% high Apr 2026 Sep 2026 mistral.ai Official leaderboard
52 MiniMax-M3.1-Flash 34% high Oct 2026 Oct 2026 huggingface.co Official leaderboard
53 Gemma 4 31B IT 33.3% on Apr 2026 Sep 2026 deepmind.google Official leaderboard
54 Llama 3.1 8B Instruct 8B 31.4% none Oct 2026 huggingface.co Official leaderboard
55 Qwen 3.8 27B 30.9% off Aug 2026 Oct 2026 huggingface.co Official leaderboard
56 Apertus 8B Instruct 29.1% none Sep 2025 Sep 2026 huggingface.co Official leaderboard
57 Gemma 4 12B 25.9% on Jun 2026 Nov 2026 huggingface.co Official leaderboard
58 Apertus 1.5 8B 25.3% on Jul 2026 Sep 2026 huggingface.co Official leaderboard
59 Gemma 4 E2B 24.3% off Mar 2026 Oct 2026 huggingface.co Official leaderboard
60 Gemma 4 E4B 24.1% off Mar 2026 Oct 2026 huggingface.co Official leaderboard
61 GLM-5.3-Flash 21.7% on Aug 2026 Sep 2026 huggingface.co Official leaderboard
62 Ripple-65B Preview 18.3% on Nov 2026 Nov 2026 huggingface.co Official leaderboard
#ModelScoreSettingReleasedRun onLinkSource
1 GPT-6.2 Sol 87.7% high Nov 2026 Nov 2026 developers.openai.com Official leaderboard
2 GPT-6.1 Terra 84.3% off Oct 2026 Oct 2026 developers.openai.com Official leaderboard
3 GPT-6 Sol 83.8% high Sep 2026 Sep 2026 developers.openai.com Official leaderboard
4 Claude Fable 5.2 82.9% low Nov 2026 Nov 2026 anthropic.com Official leaderboard
5 GPT-6 Luna 82.3% high Sep 2026 Sep 2026 developers.openai.com Official leaderboard
6 GPT-6 Astra 82.2% high Sep 2026 Sep 2026 developers.openai.com Official leaderboard
7 Gemini 4 Hydrogen 79.4% high Nov 2026 Nov 2026 ai.google.dev Official leaderboard
8 GPT-6.1 Sol 78.8% low Sep 2026 Sep 2026 developers.openai.com Official leaderboard
9 GPT-5.6 Luna 78.7% off Jul 2026 Sep 2026 developers.openai.com Official leaderboard
10 GPT-5.6 Sol 78.7% off Jul 2026 Sep 2026 developers.openai.com Official leaderboard
11 Gemini 4 Neon 78.3% low Nov 2026 Nov 2026 ai.google.dev Official leaderboard
12 Grok 4.7 77.8% low Sep 2026 Nov 2026 docs.x.ai Official leaderboard
13 Gemini 4 Carbon 75.6% low Oct 2026 Nov 2026 ai.google.dev Official leaderboard
14 DeepSeek-V4-Pro 75.6% on Apr 2026 Sep 2026 huggingface.co Official leaderboard
15 Claude Sonnet 5.5 74.9% high Sep 2026 Sep 2026 anthropic.com Official leaderboard
16 Gemini 4 Argon 74.9% high Sep 2026 Oct 2026 ai.google.dev Official leaderboard
17 GPT-5.6 Terra 74.6% off Jul 2026 Sep 2026 developers.openai.com Official leaderboard
18 Claude Opus 5.5 74.5% high Sep 2026 Sep 2026 anthropic.com Official leaderboard
19 Gemini 3.1 Pro 74.3% high Feb 2026 Sep 2026 ai.google.dev Official leaderboard
20 Claude Fable 5.1 73.7% high Sep 2026 Sep 2026 anthropic.com Official leaderboard
21 Claude Opus 5 72.3% off Jul 2026 Sep 2026 anthropic.com Official leaderboard
22 Gemini 3.7 Flash 71.4% low Aug 2026 Sep 2026 ai.google.dev Official leaderboard
23 Claude Opus 4.8 70.7% high May 2026 Sep 2026 anthropic.com Official leaderboard
24 Claude Sonnet 5 70.2% high Jun 2026 Sep 2026 anthropic.com Official leaderboard
25 Claude Haiku 5.5 69% high Oct 2026 Oct 2026 anthropic.com Official leaderboard
26 GPT-4o Mar 2025 68.6% none Mar 2025 Sep 2026 platform.openai.com Official leaderboard
27 Gemini 3.8 Flash 67.8% low Sep 2026 Sep 2026 ai.google.dev Official leaderboard
28 DeepSeek V4.1 Flash 65.6% on Sep 2026 Oct 2026 huggingface.co Official leaderboard
29 GPT-5.2 64.8% off Dec 2025 Sep 2026 platform.openai.com Official leaderboard
30 Inkling 2 Small 64% minimal Nov 2026 Nov 2026 huggingface.co Official leaderboard
31 Gemini 2.5 Pro Jun 2025 63.9% minimal Jun 2025 Sep 2026 ai.google.dev Official leaderboard
32 Qwen3.8-2.4T-A95B 62.8% on Aug 2026 Sep 2026 huggingface.co Official leaderboard
33 Gemini 3.5 Flash 62.6% high May 2026 Sep 2026 ai.google.dev Official leaderboard
34 Muse Spark Open 61.7% off Oct 2026 Oct 2026 huggingface.co Official leaderboard
35 Mistral Large 4 61% high Oct 2026 Oct 2026 mistral.ai Official leaderboard
36 Inkling 57.9% high Jul 2026 Sep 2026 huggingface.co Official leaderboard
37 Kimi K3 57.6% on Jul 2026 Sep 2026 huggingface.co Official leaderboard
38 Gemini 3.5 Flash-Lite 57.5% high Jul 2026 Sep 2026 ai.google.dev Official leaderboard
39 Mistral Large 3 55.2% none Dec 2025 Sep 2026 mistral.ai Official leaderboard
40 Gemini 2.5 Flash-Lite Sep 2025 55% high Sep 2025 Sep 2026 ai.google.dev Official leaderboard
41 MiniMax-M3 48.8% on Jun 2026 Sep 2026 huggingface.co Official leaderboard
42 GPT-4.1 nano 48.7% none Apr 2025 Sep 2026 platform.openai.com Official leaderboard
43 Mistral Small 4 45.6% off Mar 2026 Sep 2026 mistral.ai Official leaderboard
44 Inkling-Small 43.4% high Jul 2026 Sep 2026 huggingface.co Official leaderboard
45 Apertus 1.5 70B 42.6% on Jul 2026 Sep 2026 huggingface.co Official leaderboard
46 Gemma 4 26B A4B 37.5% on Apr 2026 Oct 2026 huggingface.co Official leaderboard
47 GPT-3.5 Turbo Jan 2024 36.1% none Jan 2024 Sep 2026 platform.openai.com Official leaderboard
48 Claude Haiku 4.5 35.4% off Oct 2025 Sep 2026 anthropic.com Official leaderboard
49 Mistral Medium 3.5 34.4% high Apr 2026 Sep 2026 mistral.ai Official leaderboard
50 MiniMax-M3.1-Flash 34% high Oct 2026 Oct 2026 huggingface.co Official leaderboard
51 Gemma 4 31B IT 33.3% on Apr 2026 Sep 2026 deepmind.google Official leaderboard
52 Qwen 3.8 27B 30.9% off Aug 2026 Oct 2026 huggingface.co Official leaderboard
53 Apertus 8B Instruct 29.1% none Sep 2025 Sep 2026 huggingface.co Official leaderboard
54 Gemma 4 12B 25.9% on Jun 2026 Nov 2026 huggingface.co Official leaderboard
55 Apertus 1.5 8B 25.3% on Jul 2026 Sep 2026 huggingface.co Official leaderboard
56 Gemma 4 E2B 24.3% off Mar 2026 Oct 2026 huggingface.co Official leaderboard
57 Gemma 4 E4B 24.1% off Mar 2026 Oct 2026 huggingface.co Official leaderboard
58 GLM-5.3-Flash 21.7% on Aug 2026 Sep 2026 huggingface.co Official leaderboard
59 Ripple-65B Preview 18.3% on Nov 2026 Nov 2026 huggingface.co Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s own leaderboard: Grammar and Vocabulary. Each ranks models on one part of the test. GPT-6.2 Sol leads both. Arrows show how far a model moved from its overall rank among the models we track.

Grammar

59 models we track have a score here.

  1. 1 GPT-6.2 Sol 93.3%
  2. 2 GPT-6 Astra ↑4 92.1%
  3. 3 Claude Fable 5.2 ↑1 91.4%
  4. 4 GPT-6 Sol ↓1 90%
  5. 5 GPT-5.6 Sol ↑5 89.9%
  6. 6 GPT-6.1 Sol ↑2 89.8%
  7. 7 GPT-5.6 Luna ↑2 88.1%
  8. 8 GPT-6.1 Terra ↓6 87.9%
  9. 9 Claude Opus 5.5 ↑9 86.6%
  10. 10 Gemini 4 Hydrogen ↓3 86.6%

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
GPT-6.2 Sol
low high
87.7%
GPT-6.1 Terra
off high
84.3%
GPT-6 Sol
off
83.8%
Claude Fable 5.2
low high
82.9%
GPT-6 Luna
off high
82.3%
GPT-6 Astra
low high
82.2%
Gemini 4 Hydrogen
low high
79.4%
GPT-6.1 Sol
low
78.8%
GPT-5.6 Luna
off
78.7%
GPT-5.6 Sol
off high
78.7%
Grok 4.7
low
77.8%
DeepSeek-V4-Pro
off on
75.6%

Source: Official Miðeind Icelandic Generative Quality leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-6.2 Sol high 87.7%
GPT-6.1 Terra off 84.3%
GPT-6 Sol high 83.8%
Claude Fable 5.2 low 82.9%
GPT-6 Luna high 82.3%
GPT-6 Astra high 82.2%
GPT-6 Sol off 82.2%
GPT-6.2 Sol low 80.9%
Gemini 4 Hydrogen high 79.4%
GPT-6.1 Sol low 78.8%
GPT-5.6 Luna off 78.7%
GPT-5.6 Sol off 78.7%
Gemini 4 Neon low 78.3%
GPT-5.6 Luna high 78.2%
GPT-6 Luna off 78.1%
GPT-6.1 Sol high 78.1%
Grok 4.7 low 77.8%
Gemini 4 Hydrogen low 77.5%
Claude Fable 5.2 high 77.3%
GPT-6.1 Terra high 77.3%
Grok 4.7 high 77%
GPT-5.6 Sol high 76.3%
GPT-6 Astra low 75.9%
Gemini 4 Carbon low 75.6%
DeepSeek-V4-Pro on 75.6%
Claude Sonnet 5.5 high 74.9%
Gemini 4 Argon high 74.9%
Claude Sonnet 5.5 off 74.8%
GPT-5.6 Terra off 74.6%
Claude Opus 5.5 high 74.5%
Gemini 3.1 Pro high 74.3%
GPT-5.6 Terra high 73.7%
Claude Fable 5.1 high 73.7%
Claude Opus 5 off 72.3%
Claude Opus 5 high 71.6%
Gemini 3.7 Flash low 71.4%
Claude Opus 4.8 high 70.7%
Claude Opus 4.8 off 70.5%
Gemini 4 Argon off 70.4%
Gemini 3.7 Flash high 70.3%
Claude Sonnet 5 high 70.2%
DeepSeek-V4-Pro off 69.7%
Gemini 3.1 Pro low 69.3%
Claude Haiku 5.5 high 69%
Claude Haiku 5.5 off 68.7%
GPT-4o Mar 2025 none 68.6%
Gemini 3.8 Flash low 67.8%
DeepSeek V4.1 Flash on 65.6%
DeepSeek V4.1 Flash off 65.4%
GPT-5.2 off 64.8%
GPT-5.2 high 64.3%
Inkling 2 Small minimal 64%
Gemini 2.5 Pro Jun 2025 minimal 63.9%
Gemini 2.5 Pro Jun 2025 high 62.8%
Qwen3.8-2.4T-A95B on 62.8%
Gemini 3.5 Flash high 62.6%
Gemini 3.5 Flash off 62.1%
Muse Spark Open off 61.7%
Mistral Large 4 high 61%
Muse Spark Open high 60.8%
Gemini 3.8 Flash high 60.4%
Claude Sonnet 5 off 60%
Inkling 2 Small high 59.8%
Mistral Large 4 low 58.4%
Inkling high 57.9%
Kimi K3 on 57.6%
Gemini 3.5 Flash-Lite high 57.5%
Mistral Large 3 none 55.2%
Gemini 3.5 Flash-Lite off 55.1%
Gemini 2.5 Flash-Lite Sep 2025 high 55%
Gemini 2.5 Flash-Lite Sep 2025 off 52.5%
Kimi K3 off 52.5%
Inkling off 49%
MiniMax-M3 on 48.8%
GPT-4.1 nano none 48.7%
Mistral Small 4 off 45.6%
Mistral Small 4 high 44.8%
Inkling-Small high 43.4%
Apertus 1.5 70B on 42.6%
Gemma 4 26B A4B on 37.5%
Gemma 4 26B A4B off 37.5%
GPT-3.5 Turbo Jan 2024 none 36.1%
Claude Haiku 4.5 off 35.4%
Mistral Medium 3.5 high 34.4%
Claude Haiku 4.5 high 34.2%
Mistral Medium 3.5 off 34.1%
MiniMax-M3.1-Flash high 34%
MiniMax-M3.1-Flash off 33.7%
Gemma 4 31B IT on 33.3%
Qwen 3.8 27B off 30.9%
Apertus 8B Instruct none 29.1%
Gemma 4 12B on 25.9%
Apertus 1.5 8B on 25.3%
Gemma 4 12B off 24.5%
Qwen 3.8 27B xhigh 24.4%
Gemma 4 E2B off 24.3%
Gemma 4 E4B off 24.1%
Gemma 4 E4B on 22.1%
Gemma 4 E2B on 21.9%
GLM-5.3-Flash on 21.7%
Ripple-65B Preview on 18.3%

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
GPT-6.2 Sol
low high
87.7%
GPT-6.1 Terra
off high
84.3%
GPT-6 Sol
off
83.8%
Claude Fable 5.2
low high
82.9%
GPT-6 Luna
off high
82.3%
GPT-6 Astra
low high
82.2%
Gemini 4 Hydrogen
low high
79.4%
GPT-6.1 Sol
low
78.8%
GPT-5.6 Luna
off
78.7%
GPT-5.6 Sol
off high
78.7%
Grok 4.7
low
77.8%
DeepSeek-V4-Pro
off on
75.6%

Source: Official Miðeind Icelandic Generative Quality leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-6.2 Sol high 87.7%
GPT-6.1 Terra off 84.3%
GPT-6 Sol high 83.8%
Claude Fable 5.2 low 82.9%
GPT-6 Luna high 82.3%
GPT-6 Astra high 82.2%
GPT-6 Sol off 82.2%
GPT-6.2 Sol low 80.9%
Gemini 4 Hydrogen high 79.4%
GPT-6.1 Sol low 78.8%
GPT-5.6 Luna off 78.7%
GPT-5.6 Sol off 78.7%
Gemini 4 Neon low 78.3%
GPT-5.6 Luna high 78.2%
GPT-6 Luna off 78.1%
GPT-6.1 Sol high 78.1%
Grok 4.7 low 77.8%
Gemini 4 Hydrogen low 77.5%
Claude Fable 5.2 high 77.3%
GPT-6.1 Terra high 77.3%
Grok 4.7 high 77%
GPT-5.6 Sol high 76.3%
GPT-6 Astra low 75.9%
Gemini 4 Carbon low 75.6%
DeepSeek-V4-Pro on 75.6%
Claude Sonnet 5.5 high 74.9%
Gemini 4 Argon high 74.9%
Claude Sonnet 5.5 off 74.8%
GPT-5.6 Terra off 74.6%
Claude Opus 5.5 high 74.5%
Gemini 3.1 Pro high 74.3%
GPT-5.6 Terra high 73.7%
Claude Fable 5.1 high 73.7%
Claude Opus 5 off 72.3%
Claude Opus 5 high 71.6%
Gemini 3.7 Flash low 71.4%
Claude Opus 4.8 high 70.7%
Claude Opus 4.8 off 70.5%
Gemini 4 Argon off 70.4%
Gemini 3.7 Flash high 70.3%
Claude Sonnet 5 high 70.2%
DeepSeek-V4-Pro off 69.7%
Gemini 3.1 Pro low 69.3%
Claude Haiku 5.5 high 69%
Claude Haiku 5.5 off 68.7%
GPT-4o Mar 2025 none 68.6%
Gemini 3.8 Flash low 67.8%
DeepSeek V4.1 Flash on 65.6%
DeepSeek V4.1 Flash off 65.4%
GPT-5.2 off 64.8%
GPT-5.2 high 64.3%
Inkling 2 Small minimal 64%
Gemini 2.5 Pro Jun 2025 minimal 63.9%
Gemini 2.5 Pro Jun 2025 high 62.8%
Qwen3.8-2.4T-A95B on 62.8%
Gemini 3.5 Flash high 62.6%
Gemini 3.5 Flash off 62.1%
Muse Spark Open off 61.7%
Mistral Large 4 high 61%
Muse Spark Open high 60.8%
Gemini 3.8 Flash high 60.4%
Claude Sonnet 5 off 60%
Inkling 2 Small high 59.8%
Mistral Large 4 low 58.4%
Inkling high 57.9%
Kimi K3 on 57.6%
Gemini 3.5 Flash-Lite high 57.5%
Mistral Large 3 none 55.2%
Gemini 3.5 Flash-Lite off 55.1%
Gemini 2.5 Flash-Lite Sep 2025 high 55%
Gemini 2.5 Flash-Lite Sep 2025 off 52.5%
Kimi K3 off 52.5%
Inkling off 49%
MiniMax-M3 on 48.8%
GPT-4.1 nano none 48.7%
Mistral Small 4 off 45.6%
Mistral Small 4 high 44.8%
Inkling-Small high 43.4%
Apertus 1.5 70B on 42.6%
Gemma 4 26B A4B on 37.5%
Gemma 4 26B A4B off 37.5%
GPT-3.5 Turbo Jan 2024 none 36.1%
Claude Haiku 4.5 off 35.4%
Mistral Medium 3.5 high 34.4%
Claude Haiku 4.5 high 34.2%
Mistral Medium 3.5 off 34.1%
MiniMax-M3.1-Flash high 34%
MiniMax-M3.1-Flash off 33.7%
Gemma 4 31B IT on 33.3%
Qwen 3.8 27B off 30.9%
Apertus 8B Instruct none 29.1%
Gemma 4 12B on 25.9%
Apertus 1.5 8B on 25.3%
Gemma 4 12B off 24.5%
Qwen 3.8 27B xhigh 24.4%
Gemma 4 E2B off 24.3%
Gemma 4 E4B off 24.1%
Gemma 4 E4B on 22.1%
Gemma 4 E2B on 21.9%
GLM-5.3-Flash on 21.7%
Ripple-65B Preview on 18.3%

5 models have held number one on the official leaderboard since its release in September 2026

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

OpenAI
Own leaderboard Passed 85% in Nov 2026, by GPT-6.2 Sol
GPT-6.2 Sol 87.7%

Source: Official Miðeind Icelandic Generative Quality leaderboard

GPT-6.2 Sol on the official leaderboard passed 85% in November 2026

That was 2 months after its release. By the same age, 21% of the benchmarks from 2026 had passed 85%. 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% 2024 2025 2026 2027 Near the ceiling (85%) GPT-6.2 Sol GPT-6 Astra DeepSeek-V4-Pro GPT-4o GPT-3.5 Turbo 0% 20% 40% 60% 80% 100% 2024 2025 2026 2027 Near the ceiling (85%) GPT-6.2 Sol GPT-5.6 Luna

Source: Official Miðeind Icelandic Generative Quality 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 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (62 entries).
  • Pass Models scored 59 models we track, plus 3 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Neon.
  • Pass Frontier models 9 of the top 10 have a score, led by Claude Fable 5.2 (#2).
  • Note Headroom Saturated. The best is 87.7%, past 85% since November 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on mideind-icelandic-llm-leaderboard.static.hf.space: 101 rows for models we track and 3 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