Benchmark Almanac

Uncheatable Eval

Gemma 4 31B IT 6.29 leads since April 2026
Not in model cards

Models predict text from recently published documents. The evaluation measures perplexity across sources including news, research papers, fiction and code.

Made by
Kaifeng Tan and colleagues
Released
May 2024
Kind
Reading comprehension
Level
University
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Text, Code
Interaction
Single turn
Items from
Real work and use
Run by
Anyone
In views
Kept fresh
Strongest model tested
Gemma 4 31B IT #101 on the capability index
Last model added
April 2026
Scores refreshed
November 20th, 2026
95
Entries on its own leaderboard, 33 models we track

Gemma 4 31B IT leads with 6.29 among models released by April 2026 Gemma 4 31B IT leads with 6.29 among models released by April 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 April 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemma 4 31B IT 2026-10 · Compression Ratio (%) 6.29
  2. 2 Qwen3.5-35B-A3B-Base 2026-10 · Compression Ratio (%) 6.52
  3. 3 Mistral-Small-3.1-24B-Base 2026-10 · Compression Ratio (%) 6.56
  4. 4 Gemma 4 26B A4B 2026-10 · Compression Ratio (%) 6.57
  5. 5 Gemma 3 27B PT 2026-10 · Compression Ratio (%) 6.58
  6. 6 rwkv7-g1l-13.3b-20261031 2026-10 · Compression Ratio (%) 6.60
  7. 7 Seed-OSS-36B-Base 2026-10 · Compression Ratio (%) 6.67
  8. 8 Qwen2.5-32B 2026-10 · Compression Ratio (%) 6.71
  9. 9 Ministral-3-14B-Base-2512 2026-10 · Compression Ratio (%) 6.72
  10. 10 JoyAI-LLM-Flash-Base 2026-10 · Compression Ratio (%) 6.72
  11. 11 NVIDIA-Nemotron-3-Nano-30B 2026-10 · Compression Ratio (%) 6.76
  12. 12 K2-Horizon-32B-Pretrain 2026-10 · Compression Ratio (%) 6.77
  13. 13 Qwen3-30B-A3B-Base 2026-10 · Compression Ratio (%) 6.81
  14. 14 Qwen3-14B-Base 2026-10 · Compression Ratio (%) 6.82
  15. 15 Mistral-Nemo-Base-2407 2026-10 · Compression Ratio (%) 6.84
  16. 16 Gemma 3 12B PT 2026-10 · Compression Ratio (%) 6.86
  17. 17 Qwen3.5-9B-Base 2026-10 · Compression Ratio (%) 6.87
  18. 18 granite-4.1-30b-base 2026-10 · Compression Ratio (%) 6.87
  19. 19 Qwen2.5-14B 2026-10 · Compression Ratio (%) 6.89
  20. 20 Ministral-3-8B-Base-2512 2026-10 · Compression Ratio (%) 6.97
  21. 21 rwkv7-g1l-7.2b-20261031 2026-10 · Compression Ratio (%) 7.01
  22. 22 Llama 3.1-8B 2026-10 · Compression Ratio (%) 7.08
  23. 23 Olmo-3-1125-32B 2026-10 · Compression Ratio (%) 7.08
  24. 24 Qwen3-8B-Base 2026-10 · Compression Ratio (%) 7.10
  25. 25 marin-32b-base 2026-10 · Compression Ratio (%) 7.16
  26. 26 Gemma 4 E4B 2026-10 · Compression Ratio (%) 7.19
  27. 27 Spark-X2.6-4B-Base 2026-10 · Compression Ratio (%) 7.19
  28. 28 Falcon-H1-7B-Base 2026-10 · Compression Ratio (%) 7.23
  29. 29 Qwen3.5-4B-Base 2026-10 · Compression Ratio (%) 7.28
  30. 30 Mistral 7B v0.1 2026-10 · Compression Ratio (%) 7.31
  31. 31 granite-4.1-8b-base 2026-10 · Compression Ratio (%) 7.31
  32. 32 Qwen2.5-7B 2026-10 · Compression Ratio (%) 7.34
  33. 33 K2-Horizon-7B-Pretrain 2026-10 · Compression Ratio (%) 7.43
  34. 34 marin-8b-base 2026-10 · Compression Ratio (%) 7.44
  35. 35 rwkv7-g1l-2.9b-20261031 2026-10 · Compression Ratio (%) 7.45
  36. 36 Gravity-16B-A3B-Base 2026-10 · Compression Ratio (%) 7.46
  37. 37 Qwen3-4B-Base 2026-10 · Compression Ratio (%) 7.46
  38. 38 Ministral-3-3B-Base-2512 2026-10 · Compression Ratio (%) 7.47
  39. 39 Nanbeige4.2-3B-Base 2026-10 · Compression Ratio (%) 7.51
  40. 40 Hunyuan-7B-Pretrain 2026-10 · Compression Ratio (%) 7.51
  41. 41 Llama-2-13b-hf 2026-10 · Compression Ratio (%) 7.53
  42. 42 Gemma 3 4B PT 2026-10 · Compression Ratio (%) 7.53
  43. 43 Apertus-8B-2509 2026-10 · Compression Ratio (%) 7.56
  44. 44 Minitron-8B-Base 2026-10 · Compression Ratio (%) 7.62
  45. 45 Llama 3.2 3B 2026-10 · Compression Ratio (%) 7.69
  46. 46 Olmo-Hybrid-7B 2026-10 · Compression Ratio (%) 7.70
  47. 47 Zamba2-7B 2026-10 · Compression Ratio (%) 7.73
  48. 48 K2-Horizon-3.7B-Pretrain 2026-10 · Compression Ratio (%) 7.78
  49. 49 granite-4.1-3b-base 2026-10 · Compression Ratio (%) 7.81
  50. 50 Gemma 4 E2B 2026-10 · Compression Ratio (%) 7.83
  51. 51 ZAYA1-base 2026-10 · Compression Ratio (%) 7.83
  52. 52 Qwen2.5-3B 2026-10 · Compression Ratio (%) 7.84
  53. 53 Nanbeige4-3B-Base 2026-10 · Compression Ratio (%) 7.84
  54. 54 Olmo-3-1025-7B 2026-10 · Compression Ratio (%) 7.85
  55. 55 ZGCM-1-7B-Pretrain 2026-10 · Compression Ratio (%) 7.86
  56. 56 Falcon-H1-3B-Base 2026-10 · Compression Ratio (%) 7.87
  57. 57 Falcon3-7B-Base 2026-10 · Compression Ratio (%) 7.88
  58. 58 Llama-2-7b-hf 2026-10 · Compression Ratio (%) 7.91
  59. 59 SmolLM3-3B-Base 2026-10 · Compression Ratio (%) 7.98
  60. 60 MiniCPM5-2B-Base 2026-10 · Compression Ratio (%) 7.99
  61. 61 rwkv7-g1l-1.5b-20261031 2026-10 · Compression Ratio (%) 7.99
  62. 62 Llama-3.1-Minitron-4B-Width 2026-10 · Compression Ratio (%) 8.01
  63. 63 gemma-2-2b 2026-10 · Compression Ratio (%) 8.05
  64. 64 Qwen3.5-2B-Base 2026-10 · Compression Ratio (%) 8.06
  65. 65 Qwen3-1.7B-Base 2026-10 · Compression Ratio (%) 8.08
  66. 66 Spark-X2.5-1.7B-Base 2026-10 · Compression Ratio (%) 8.11
  67. 67 Minitron-4B-Base 2026-10 · Compression Ratio (%) 8.15
  68. 68 stablelm-3b-4e1t 2026-10 · Compression Ratio (%) 8.18
  69. 69 Qwen2.5-1.5B 2026-10 · Compression Ratio (%) 8.31
  70. 70 Index-1.9B 2026-10 · Compression Ratio (%) 8.33
  71. 71 Youtu-LLM-2B-Base 2026-10 · Compression Ratio (%) 8.39
  72. 72 Llama-3.1-Minitron-4B-Depth 2026-10 · Compression Ratio (%) 8.44
  73. 73 Falcon-H1-1.5B-Deep-Base 2026-10 · Compression Ratio (%) 8.47
  74. 74 Llama 3.2 1B 2026-10 · Compression Ratio (%) 8.51
  75. 75 LFM2.5-8B-A1B-Base 2026-10 · Compression Ratio (%) 8.65
  76. 76 Falcon-H1-1.5B-Base 2026-10 · Compression Ratio (%) 8.65
  77. 77 LFM3-1.7B-Base 2026-10 · Compression Ratio (%) 8.71
  78. 78 SmolLM2-1.7B 2026-10 · Compression Ratio (%) 8.71
  79. 79 stablelm-2-1_6b 2026-10 · Compression Ratio (%) 8.74
  80. 80 Zamba2-2.7B 2026-10 · Compression Ratio (%) 8.82
  81. 81 K2-Horizon-0.9B-Pretrain 2026-10 · Compression Ratio (%) 8.82
  82. 82 AFM-4.5B-Base-KDA-Only 2026-10 · Compression Ratio (%) 8.86
  83. 83 Qwen3.5-0.8B-Base 2026-10 · Compression Ratio (%) 8.94
  84. 84 Qwen3-0.6B-Base 2026-10 · Compression Ratio (%) 9.05
  85. 85 MiniCPM5-1B-Base 2026-10 · Compression Ratio (%) 9.07
  86. 86 Gemma 3 1B PT 2026-10 · Compression Ratio (%) 9.32
  87. 87 rwkv7-g1d-0.4b-20260210 2026-10 · Compression Ratio (%) 9.35
  88. 88 Qwen2.5-0.5B 2026-10 · Compression Ratio (%) 9.54
  89. 89 ERNIE-4.5-0.3B-Base-PT 2026-10 · Compression Ratio (%) 10.1
  90. 90 SmolLM2-360M 2026-10 · Compression Ratio (%) 10.1
  91. 91 rwkv7a-g1d-0.1b-20260212 2026-10 · Compression Ratio (%) 10.2
  92. 92 Falcon-H1-0.5B-Base 2026-10 · Compression Ratio (%) 10.2
  93. 93 rwkv7-g1d-0.1b-20260129 2026-10 · Compression Ratio (%) 10.6
  94. 94 Gemma 3 270M 2026-10 · Compression Ratio (%) 11.1
  95. 95 SmolLM2-135M 2026-10 · Compression Ratio (%) 11.5

Source: Official Uncheatable Eval leaderboard

  1. 1 Gemma 4 31B IT 2026-10 · Compression Ratio (%) 6.29
  2. 2 Qwen3.5-35B-A3B-Base 2026-10 · Compression Ratio (%) 6.52
  3. 3 Gemma 4 26B A4B 2026-10 · Compression Ratio (%) 6.57
  4. 4 Gemma 3 27B PT 2026-10 · Compression Ratio (%) 6.58
  5. 5 Seed-OSS-36B-Base 2026-10 · Compression Ratio (%) 6.67
  6. 6 Qwen2.5-32B 2026-10 · Compression Ratio (%) 6.71
  7. 7 Qwen3-30B-A3B-Base 2026-10 · Compression Ratio (%) 6.81
  8. 8 Qwen3-14B-Base 2026-10 · Compression Ratio (%) 6.82
  9. 9 Mistral-Nemo-Base-2407 2026-10 · Compression Ratio (%) 6.84
  10. 10 Gemma 3 12B PT 2026-10 · Compression Ratio (%) 6.86
  11. 11 Qwen3.5-9B-Base 2026-10 · Compression Ratio (%) 6.87
  12. 12 Qwen2.5-14B 2026-10 · Compression Ratio (%) 6.89
  13. 13 Llama 3.1-8B 2026-10 · Compression Ratio (%) 7.08
  14. 14 Qwen3-8B-Base 2026-10 · Compression Ratio (%) 7.10
  15. 15 Gemma 4 E4B 2026-10 · Compression Ratio (%) 7.19
  16. 16 Qwen3.5-4B-Base 2026-10 · Compression Ratio (%) 7.28
  17. 17 Mistral 7B v0.1 2026-10 · Compression Ratio (%) 7.31
  18. 18 Qwen2.5-7B 2026-10 · Compression Ratio (%) 7.34
  19. 19 Qwen3-4B-Base 2026-10 · Compression Ratio (%) 7.46
  20. 20 Gemma 3 4B PT 2026-10 · Compression Ratio (%) 7.53
  21. 21 Llama 3.2 3B 2026-10 · Compression Ratio (%) 7.69
  22. 22 Gemma 4 E2B 2026-10 · Compression Ratio (%) 7.83
  23. 23 Qwen2.5-3B 2026-10 · Compression Ratio (%) 7.84
  24. 24 Qwen3.5-2B-Base 2026-10 · Compression Ratio (%) 8.06
  25. 25 Qwen2.5-1.5B 2026-10 · Compression Ratio (%) 8.31
  26. 26 Llama 3.2 1B 2026-10 · Compression Ratio (%) 8.51
  27. 27 SmolLM2-1.7B 2026-10 · Compression Ratio (%) 8.71
  28. 28 Qwen3.5-0.8B-Base 2026-10 · Compression Ratio (%) 8.94
  29. 29 Gemma 3 1B PT 2026-10 · Compression Ratio (%) 9.32
  30. 30 Qwen2.5-0.5B 2026-10 · Compression Ratio (%) 9.54
  31. 31 SmolLM2-360M 2026-10 · Compression Ratio (%) 10.1
  32. 32 Gemma 3 270M 2026-10 · Compression Ratio (%) 11.1
  33. 33 SmolLM2-135M 2026-10 · Compression Ratio (%) 11.5

Source: Official Uncheatable Eval leaderboard

#EntryScoreSettingDateSource
1 Gemma 4 31B IT 6.29 2026-10 · Compression Ratio (%) Apr 2026 Official leaderboard
2 Qwen3.5-35B-A3B-Base 6.52 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
3 Mistral-Small-3.1-24B-Base 24B 6.56 2026-10 · Compression Ratio (%) Official leaderboard
4 Gemma 4 26B A4B 6.57 2026-10 · Compression Ratio (%) Apr 2026 Official leaderboard
5 Gemma 3 27B PT 6.58 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
6 rwkv7-g1l-13.3b-20261031 13.3B 6.60 2026-10 · Compression Ratio (%) Official leaderboard
7 Seed-OSS-36B-Base 6.67 2026-10 · Compression Ratio (%) Aug 2025 Official leaderboard
8 Qwen2.5-32B 6.71 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
9 Ministral-3-14B-Base-2512 14B 6.72 2026-10 · Compression Ratio (%) Official leaderboard
10 JoyAI-LLM-Flash-Base 48.9B 6.72 2026-10 · Compression Ratio (%) Official leaderboard
11 NVIDIA-Nemotron-3-Nano-30B 31.6B 6.76 2026-10 · Compression Ratio (%) Official leaderboard
12 K2-Horizon-32B-Pretrain 34.8B 6.77 2026-10 · Compression Ratio (%) Official leaderboard
13 Qwen3-30B-A3B-Base 6.81 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
14 Qwen3-14B-Base 6.82 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
15 Mistral-Nemo-Base-2407 6.84 2026-10 · Compression Ratio (%) Jul 2024 Official leaderboard
16 Gemma 3 12B PT 6.86 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
17 Qwen3.5-9B-Base 6.87 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
18 granite-4.1-30b-base 28.9B 6.87 2026-10 · Compression Ratio (%) Official leaderboard
19 Qwen2.5-14B 6.89 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
20 Ministral-3-8B-Base-2512 8.9B 6.97 2026-10 · Compression Ratio (%) Official leaderboard
21 rwkv7-g1l-7.2b-20261031 7.2B 7.01 2026-10 · Compression Ratio (%) Official leaderboard
22 Llama 3.1-8B 7.08 2026-10 · Compression Ratio (%) Jul 2024 Official leaderboard
23 Olmo-3-1125-32B 32.2B 7.08 2026-10 · Compression Ratio (%) Official leaderboard
24 Qwen3-8B-Base 7.10 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
25 marin-32b-base 32.5B 7.16 2026-10 · Compression Ratio (%) Official leaderboard
26 Gemma 4 E4B 7.19 2026-10 · Compression Ratio (%) Mar 2026 Official leaderboard
27 Spark-X2.6-4B-Base 4.1B 7.19 2026-10 · Compression Ratio (%) Official leaderboard
28 Falcon-H1-7B-Base 7.6B 7.23 2026-10 · Compression Ratio (%) Official leaderboard
29 Qwen3.5-4B-Base 7.28 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
30 Mistral 7B v0.1 7.31 2026-10 · Compression Ratio (%) Sep 2023 Official leaderboard
31 granite-4.1-8b-base 8.4B 7.31 2026-10 · Compression Ratio (%) Official leaderboard
32 Qwen2.5-7B 7.34 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
33 K2-Horizon-7B-Pretrain 9B 7.43 2026-10 · Compression Ratio (%) Official leaderboard
34 marin-8b-base 8B 7.44 2026-10 · Compression Ratio (%) Official leaderboard
35 rwkv7-g1l-2.9b-20261031 2.9B 7.45 2026-10 · Compression Ratio (%) Official leaderboard
36 Gravity-16B-A3B-Base 16.2B 7.46 2026-10 · Compression Ratio (%) Official leaderboard
37 Qwen3-4B-Base 7.46 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
38 Ministral-3-3B-Base-2512 3.8B 7.47 2026-10 · Compression Ratio (%) Official leaderboard
39 Nanbeige4.2-3B-Base 4.2B 7.51 2026-10 · Compression Ratio (%) Official leaderboard
40 Hunyuan-7B-Pretrain 7.5B 7.51 2026-10 · Compression Ratio (%) Official leaderboard
41 Llama-2-13b-hf 13B 7.53 2026-10 · Compression Ratio (%) Official leaderboard
42 Gemma 3 4B PT 7.53 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
43 Apertus-8B-2509 8.1B 7.56 2026-10 · Compression Ratio (%) Official leaderboard
44 Minitron-8B-Base 8.3B 7.62 2026-10 · Compression Ratio (%) Official leaderboard
45 Llama 3.2 3B 7.69 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
46 Olmo-Hybrid-7B 7.4B 7.70 2026-10 · Compression Ratio (%) Official leaderboard
47 Zamba2-7B 7.4B 7.73 2026-10 · Compression Ratio (%) Official leaderboard
48 K2-Horizon-3.7B-Pretrain 5.1B 7.78 2026-10 · Compression Ratio (%) Official leaderboard
49 granite-4.1-3b-base 3.4B 7.81 2026-10 · Compression Ratio (%) Official leaderboard
50 Gemma 4 E2B 7.83 2026-10 · Compression Ratio (%) Mar 2026 Official leaderboard
51 ZAYA1-base 8.8B 7.83 2026-10 · Compression Ratio (%) Official leaderboard
52 Qwen2.5-3B 7.84 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
53 Nanbeige4-3B-Base 3.9B 7.84 2026-10 · Compression Ratio (%) Official leaderboard
54 Olmo-3-1025-7B 7.3B 7.85 2026-10 · Compression Ratio (%) Official leaderboard
55 ZGCM-1-7B-Pretrain 7.4B 7.86 2026-10 · Compression Ratio (%) Official leaderboard
56 Falcon-H1-3B-Base 3.1B 7.87 2026-10 · Compression Ratio (%) Official leaderboard
57 Falcon3-7B-Base 7.5B 7.88 2026-10 · Compression Ratio (%) Official leaderboard
58 Llama-2-7b-hf 6.7B 7.91 2026-10 · Compression Ratio (%) Official leaderboard
59 SmolLM3-3B-Base 3.1B 7.98 2026-10 · Compression Ratio (%) Official leaderboard
60 MiniCPM5-2B-Base 2.5B 7.99 2026-10 · Compression Ratio (%) Official leaderboard
61 rwkv7-g1l-1.5b-20261031 1.5B 7.99 2026-10 · Compression Ratio (%) Official leaderboard
62 Llama-3.1-Minitron-4B-Width 4.5B 8.01 2026-10 · Compression Ratio (%) Official leaderboard
63 gemma-2-2b 2.6B 8.05 2026-10 · Compression Ratio (%) Official leaderboard
64 Qwen3.5-2B-Base 8.06 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
65 Qwen3-1.7B-Base 1.7B 8.08 2026-10 · Compression Ratio (%) Official leaderboard
66 Spark-X2.5-1.7B-Base 1.7B 8.11 2026-10 · Compression Ratio (%) Official leaderboard
67 Minitron-4B-Base 4.2B 8.15 2026-10 · Compression Ratio (%) Official leaderboard
68 stablelm-3b-4e1t 2.8B 8.18 2026-10 · Compression Ratio (%) Official leaderboard
69 Qwen2.5-1.5B 8.31 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
70 Index-1.9B 2.2B 8.33 2026-10 · Compression Ratio (%) Official leaderboard
71 Youtu-LLM-2B-Base 2B 8.39 2026-10 · Compression Ratio (%) Official leaderboard
72 Llama-3.1-Minitron-4B-Depth 4.5B 8.44 2026-10 · Compression Ratio (%) Official leaderboard
73 Falcon-H1-1.5B-Deep-Base 1.6B 8.47 2026-10 · Compression Ratio (%) Official leaderboard
74 Llama 3.2 1B 8.51 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
75 LFM2.5-8B-A1B-Base 8.5B 8.65 2026-10 · Compression Ratio (%) Official leaderboard
76 Falcon-H1-1.5B-Base 1.6B 8.65 2026-10 · Compression Ratio (%) Official leaderboard
77 LFM3-1.7B-Base 1.7B 8.71 2026-10 · Compression Ratio (%) Official leaderboard
78 SmolLM2-1.7B 8.71 2026-10 · Compression Ratio (%) Oct 2024 Official leaderboard
79 stablelm-2-1_6b 1.6B 8.74 2026-10 · Compression Ratio (%) Official leaderboard
80 Zamba2-2.7B 2.7B 8.82 2026-10 · Compression Ratio (%) Official leaderboard
81 K2-Horizon-0.9B-Pretrain 1.1B 8.82 2026-10 · Compression Ratio (%) Official leaderboard
82 AFM-4.5B-Base-KDA-Only 5B 8.86 2026-10 · Compression Ratio (%) Official leaderboard
83 Qwen3.5-0.8B-Base 8.94 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
84 Qwen3-0.6B-Base 0.6B 9.05 2026-10 · Compression Ratio (%) Official leaderboard
85 MiniCPM5-1B-Base 1.1B 9.07 2026-10 · Compression Ratio (%) Official leaderboard
86 Gemma 3 1B PT 9.32 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
87 rwkv7-g1d-0.4b-20260210 0.5B 9.35 2026-10 · Compression Ratio (%) Official leaderboard
88 Qwen2.5-0.5B 9.54 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
89 ERNIE-4.5-0.3B-Base-PT 0.4B 10.1 2026-10 · Compression Ratio (%) Official leaderboard
90 SmolLM2-360M 10.1 2026-10 · Compression Ratio (%) Oct 2024 Official leaderboard
91 rwkv7a-g1d-0.1b-20260212 1B 10.2 2026-10 · Compression Ratio (%) Official leaderboard
92 Falcon-H1-0.5B-Base 0.5B 10.2 2026-10 · Compression Ratio (%) Official leaderboard
93 rwkv7-g1d-0.1b-20260129 0.2B 10.6 2026-10 · Compression Ratio (%) Official leaderboard
94 Gemma 3 270M 11.1 2026-10 · Compression Ratio (%) Aug 2025 Official leaderboard
95 SmolLM2-135M 11.5 2026-10 · Compression Ratio (%) Oct 2024 Official leaderboard
#ModelScoreSettingDateSource
1 Gemma 4 31B IT 6.29 2026-10 · Compression Ratio (%) Apr 2026 Official leaderboard
2 Qwen3.5-35B-A3B-Base 6.52 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
3 Gemma 4 26B A4B 6.57 2026-10 · Compression Ratio (%) Apr 2026 Official leaderboard
4 Gemma 3 27B PT 6.58 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
5 Seed-OSS-36B-Base 6.67 2026-10 · Compression Ratio (%) Aug 2025 Official leaderboard
6 Qwen2.5-32B 6.71 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
7 Qwen3-30B-A3B-Base 6.81 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
8 Qwen3-14B-Base 6.82 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
9 Mistral-Nemo-Base-2407 6.84 2026-10 · Compression Ratio (%) Jul 2024 Official leaderboard
10 Gemma 3 12B PT 6.86 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
11 Qwen3.5-9B-Base 6.87 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
12 Qwen2.5-14B 6.89 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
13 Llama 3.1-8B 7.08 2026-10 · Compression Ratio (%) Jul 2024 Official leaderboard
14 Qwen3-8B-Base 7.10 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
15 Gemma 4 E4B 7.19 2026-10 · Compression Ratio (%) Mar 2026 Official leaderboard
16 Qwen3.5-4B-Base 7.28 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
17 Mistral 7B v0.1 7.31 2026-10 · Compression Ratio (%) Sep 2023 Official leaderboard
18 Qwen2.5-7B 7.34 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
19 Qwen3-4B-Base 7.46 2026-10 · Compression Ratio (%) Apr 2025 Official leaderboard
20 Gemma 3 4B PT 7.53 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
21 Llama 3.2 3B 7.69 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
22 Gemma 4 E2B 7.83 2026-10 · Compression Ratio (%) Mar 2026 Official leaderboard
23 Qwen2.5-3B 7.84 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
24 Qwen3.5-2B-Base 8.06 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
25 Qwen2.5-1.5B 8.31 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
26 Llama 3.2 1B 8.51 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
27 SmolLM2-1.7B 8.71 2026-10 · Compression Ratio (%) Oct 2024 Official leaderboard
28 Qwen3.5-0.8B-Base 8.94 2026-10 · Compression Ratio (%) Feb 2026 Official leaderboard
29 Gemma 3 1B PT 9.32 2026-10 · Compression Ratio (%) Mar 2025 Official leaderboard
30 Qwen2.5-0.5B 9.54 2026-10 · Compression Ratio (%) Sep 2024 Official leaderboard
31 SmolLM2-360M 10.1 2026-10 · Compression Ratio (%) Oct 2024 Official leaderboard
32 Gemma 3 270M 11.1 2026-10 · Compression Ratio (%) Aug 2025 Official leaderboard
33 SmolLM2-135M 11.5 2026-10 · Compression Ratio (%) Oct 2024 Official leaderboard

Rankings by part

We read 14 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as github cpp, github javascript and github python. Gemma 4 31B IT leads 10 of them, and Qwen3.5-35B-A3B-Base leads the other 4. Arrows show how far a model moved from its overall rank among the models we track.

github cpp

33 models we track have a score here.

  1. 1 Gemma 4 31B IT 3.08
  2. 2 Seed-OSS-36B-Base ↑3 3.11
  3. 3 Gemma 4 26B A4B 3.20
  4. 4 Qwen3.5-35B-A3B-Base ↓2 3.26
  5. 5 Qwen2.5-32B ↑1 3.27
  6. 6 Qwen3-30B-A3B-Base ↑1 3.29
  7. 7 Qwen3-14B-Base ↑1 3.31
  8. 8 Gemma 3 27B PT ↓4 3.32
  9. 9 Qwen2.5-14B ↑3 3.43
  10. 10 Qwen3-8B-Base ↑4 3.49

6 models have held number one on the official leaderboard since its release in May 2024

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

Mistral AI Qwen Google DeepMind
Own leaderboard Not tested since Apr 2026

Source: Official Uncheatable Eval 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
6.00 8.00 10.0 12.0 2024 2025 2026 6.00 8.00 10.0 12.0 2024 2025 2026

Source: Official Uncheatable Eval 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 Knowledge

All 332 in Knowledge
  • MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
  • MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. Seed1.8 92.3%
  • SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy. Qwen 3.6 Max (Preview) 73.9%
  • WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1. Tencent Hy4 preview 83.9%
  • BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly. Gemma 4 31B IT 74.4%
  • HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 7 months ago, in April 2026, for Gemma 4 26B A4B. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (95 entries).
  • Pass Models scored 33 models we track, plus 62 other entries on its leaderboard.
  • Warning Newest score The newest came 7 months ago, in April 2026, for Gemma 4 26B A4B.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Gemma 4 31B IT (#101).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 33 rows for models we track and 62 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