Benchmark Almanac

RuLES

GPT-4 Turbo Nov 2023 9.54% leads since November 2023
Not in model cards
Version 1.0

Models follow explicit rules while talking with a user. Each scenario checks automatically whether the model broke a rule.

The authors report results for open and proprietary models. Their repository has 259 GitHub stars.

Made by
Norman Mu and collaborators
Released
November 2023
Kind
Following precise instructions
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Multi-turn
Items from
Generated
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Claude 3 Opus #173 on the capability index
Last model added
February 2024
Scores refreshed
November 21st, 2026
134
Entries in its paper, 30 models we track

GPT-4 Turbo leads with 9.54% among models released by February 2024 GPT-4 Turbo leads with 9.54% among models released by February 2024

Best score per entry from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. 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.

The newest model with a score here came out in February 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

134 entries. Scroll sideways for the rest.

  1. 9.54 GPT-4 Turbo
  2. 9.53 GPT-4 Turbo (Jan 2024)
  3. 8.51 Claude 3 Opus
  4. 8.30 GPT-3.5 Turbo
  5. 7.90 GPT-3.5 Turbo
  6. 7.88 claude-instant-v1.2
  7. 7.02 Qwen1.5-72B
  8. 6.87 Claude 2.1
  9. 6.86 Yi-34B-200K-AEZAKMI-v2
  10. 6.80 GPT-3.5 Turbo
  11. 6.79 Claude 2
  12. 6.76 tulu-2-70b
  13. 6.64 dolphin-2.2-70b
  14. 6.51 Gemini 1.0 Pro
  15. 6.50 Mixtral-SlimOrca-8x7B
  16. 6.49 Instruct_Mixtral-8x7B-v0.1_Dol[..]
  17. 6.40 Nous-Hermes-2-Yi-34B
  18. 6.39 DeepSeek LLM 67B
  19. 6.39 Claude 3 Sonnet
  20. 6.38 dolphin-2.2-yi-34b-200k
  21. 6.38 Yi-34B-AEZAKMI-v1
  22. 6.30 yi-34B-v2
  23. 6.25 Nous-Hermes-2-Mixtral-8x7B-SFT
  24. 6.24 yi-34B-v3
  25. 6.15 OrionStar-Yi-34B-Chat
  26. 6.13 blossom-v4-yi-34b
  27. 6.12 deepseek-llm-67b-base
  28. 6.11 Llama-2-7b-hf
  29. 6.08 Yi-34B-200K-DARE-merge-v5
  30. 6.06 Capybara-Tess-Yi-34B-200K
  31. 6.03 platypus-yi-34b
  32. 5.99 openchat-3.5-0106
  33. 5.96 notux-8x7b-v1
  34. 5.96 orca_mini_v3_13b
  35. 5.94 SOLAR-0-70b-16bit
  36. 5.92 Mistral-7B-AEZAKMI-v2
  37. 5.87 samantha-mistral-7b
  38. 5.79 Gemma 7B
  39. 5.78 GPT-3.5 Turbo (Mar 2023)
  40. 5.76 vicuna-7b-v1.5
  41. 5.74 openchat-3.5-1210
  42. 5.73 dolphin-2.6-mixtral-8x7b
  43. 5.73 neural-chat-7b-v3-2
  44. 5.69 blossom-v3-mistral-7b
  45. 5.69 tulu-2-13b
  46. 5.66 dolphin-2.2.1-mistral-7b
  47. 5.64 CaPlatTessDolXaBoros-Yi-34B-20[..]
  48. 5.63 orca_mini_v3_70b
  49. 5.63 Mistral 7B v0.1
  50. 5.62 dolphin-2.5-mixtral-8x7b
  51. 5.62 OpenHermes-2.5-Mistral-7B
  52. 5.62 Nous-Hermes-2-Mixtral-8x7B-DPO
  53. 5.56 Samantha-1.11-70b
  54. 5.55 Orca-2-13b
  55. 5.54 blossom-v3_1-mistral-7b
  56. 5.54 Nous-Hermes-2-Llama-2-70B
  57. 5.53 Llama-2-70b-hf
  58. 5.50 Yi-34B
  59. 5.49 neural-chat-7b-v3-1
  60. 5.48 blossom-v4-mistral-7b
  61. 5.46 dolphin-2.7-mixtral-8x7b
  62. 5.45 orca_mini_v3_7b
  63. 5.43 deepseek-llm-7b-base
  64. 5.43 dolphin-2.6-mistral-7b-dpo
  65. 5.41 Mistral-7B-OpenOrca
  66. 5.40 dolphin-2.6-mistral-7b
  67. 5.40 Tulu 2 DPO 70B
  68. 5.39 dolphin-2.6-mistral-7b-dpo-las[..]
  69. 5.38 Llama-2-13b-hf
  70. 5.32 deepseek-llm-7b-chat
  71. 5.32 Qwen1.5-14B-Chat
  72. 5.25 Orca-2-7b
  73. 5.25 vicuna-13b-v1.5
  74. 5.23 mistral-7b-sft-alpha
  75. 5.19 Starling-LM-7B-alpha
  76. 5.16 Mixtral 8x7B
  77. 5.15 OpenHermes-Mixtral-8x7B
  78. 5.14 Mistral-7B-AEZAKMI-v1
  79. 5.07 Qwen1.5-7B
  80. 5.06 samantha-mistral-instruct-7b
  81. 5.06 Gemma 2B
  82. 5.05 Nous-Capybara-7B-V1.9
  83. 5.04 OpenHermes-7B
  84. 5.03 tulu-2-7b
  85. 5.03 Qwen1.5-4B-Chat
  86. 5.00 Nous-Capybara-7B-V1
  87. 5.00 OpenHermes-2-Mistral-7B
  88. 4.99 WizardLM-70B-V1.0
  89. 4.98 Qwen1.5-14B
  90. 4.95 miqu-1-70b-sf
  91. 4.90 Mistral-7B-Instruct-v0.1
  92. 4.77 neural-chat-7b-v3-3
  93. 4.74 WizardLM-13B-V1.2
  94. 4.66 Zephyr 7B Alpha
  95. 4.65 Toppy-M-7B
  96. 4.64 mythalion-13b
  97. 4.64 Xwin-LM-70B-V0.1
  98. 4.62 Llama-2-70b-chat-hf
  99. 4.57 Nous-Hermes-Llama2-13b
  100. 4.54 neural-chat-7b-v3-3-Slerp
  101. 4.53 Platypus2-70B-instruct
  102. 4.48 ReMM-v2.2-L2-13B
  103. 4.47 Llama-2-7B-32K-Instruct
  104. 4.46 mixtral-megamerge-dare-8x7b-v2
  105. 4.44 Llama-2-13b-chat-hf
  106. 4.44 ReMM-SLERP-L2-13B
  107. 4.43 firefly-mixtral-8x7b
  108. 4.43 MythoMax-L2-13b
  109. 4.40 mistral-7b-sft-beta
  110. 4.40 tulu-2-dpo-13b
  111. 4.37 Qwen1.5-7B-Chat
  112. 4.36 NeuralMonarch-7B
  113. 4.33 Llama-2-7b-chat-hf
  114. 4.33 Nous-Hermes-llama-2-7b
  115. 4.31 gemma-2b-it
  116. 4.30 samantha-1.2-mistral-7b
  117. 4.25 pygmalion-2-7b
  118. 4.24 AlphaMonarch-7B
  119. 4.24 yi-34B-v4
  120. 4.24 Zephyr 7B Beta
  121. 4.21 pygmalion-2-13b
  122. 4.04 Qwen1.5-1.8B-Chat
  123. 4.02 Qwen1.5-1.8B
  124. 3.95 Qwen1.5-0.5B-Chat
  125. 3.94 Mixtral-8x7B-Instruct-v0.1
  126. 3.78 tulu-2-dpo-7b
  127. 3.76 Mistral-7B-Instruct-v0.2
  128. 3.75 notus-7b-v1
  129. 3.75 Xwin-LM-7B-V0.2
  130. 3.65 gemma-7b-it
  131. 3.60 Xwin-LM-13B-V0.2
  132. 3.59 Qwen1.5-4B
  133. 3.52 Qwen1.5-0.5B
  134. 3.11 Yi-34B-Chat
  1. 1 GPT-4 Turbo Nov 2023, scenario instruction 9.54%
  2. 2 GPT-4 Turbo (Jan 2024) scenario instruction 9.53%
  3. 3 Claude 3 Opus 8.51%
  4. 4 GPT-3.5 Turbo Nov 2023, scenario instruction 8.30%
  5. 5 GPT-3.5 Turbo Jan 2024, scenario instruction 7.90%
  6. 6 claude-instant-v1.2 7.88%
  7. 7 Qwen1.5-72B 7.02%
  8. 8 Claude 2.1 6.87%
  9. 9 Yi-34B-200K-AEZAKMI-v2 6.86%
  10. 10 GPT-3.5 Turbo Jun 2023 6.80%
  11. 11 Claude 2 6.79%
  12. 12 tulu-2-70b 6.76%
  13. 13 dolphin-2.2-70b 6.64%
  14. 14 Gemini 1.0 Pro 6.51%
  15. 15 Mixtral-SlimOrca-8x7B 6.50%
  16. 16 Instruct_Mixtral-8x7B-v0.1_Dol[..] 6.49%
  17. 17 Nous-Hermes-2-Yi-34B 6.40%
  18. 18 DeepSeek LLM 67B 6.39%
  19. 19 Claude 3 Sonnet scenario instruction 6.39%
  20. 20 dolphin-2.2-yi-34b-200k 6.38%
  21. 21 Yi-34B-AEZAKMI-v1 6.38%
  22. 22 yi-34B-v2 6.30%
  23. 23 Nous-Hermes-2-Mixtral-8x7B-SFT 6.25%
  24. 24 yi-34B-v3 6.24%
  25. 25 OrionStar-Yi-34B-Chat 6.15%
  26. 26 blossom-v4-yi-34b 6.13%
  27. 27 deepseek-llm-67b-base 6.12%
  28. 28 Llama-2-7b-hf 6.11%
  29. 29 Yi-34B-200K-DARE-merge-v5 6.08%
  30. 30 Capybara-Tess-Yi-34B-200K 6.06%
  31. 31 platypus-yi-34b 6.03%
  32. 32 openchat-3.5-0106 5.99%
  33. 33 notux-8x7b-v1 5.96%
  34. 34 orca_mini_v3_13b 5.96%
  35. 35 SOLAR-0-70b-16bit 5.94%
  36. 36 Mistral-7B-AEZAKMI-v2 5.92%
  37. 37 samantha-mistral-7b 5.87%
  38. 38 Gemma 7B 5.79%
  39. 39 GPT-3.5 Turbo (Mar 2023) 5.78%
  40. 40 vicuna-7b-v1.5 5.76%
  41. 41 openchat-3.5-1210 5.74%
  42. 42 dolphin-2.6-mixtral-8x7b 5.73%
  43. 43 neural-chat-7b-v3-2 5.73%
  44. 44 blossom-v3-mistral-7b 5.69%
  45. 45 tulu-2-13b 5.69%
  46. 46 dolphin-2.2.1-mistral-7b 5.66%
  47. 47 CaPlatTessDolXaBoros-Yi-34B-20[..] 5.64%
  48. 48 orca_mini_v3_70b 5.63%
  49. 49 Mistral 7B v0.1 5.63%
  50. 50 dolphin-2.5-mixtral-8x7b 5.62%
  51. 51 OpenHermes-2.5-Mistral-7B 5.62%
  52. 52 Nous-Hermes-2-Mixtral-8x7B-DPO 5.62%
  53. 53 Samantha-1.11-70b 5.56%
  54. 54 Orca-2-13b 5.55%
  55. 55 blossom-v3_1-mistral-7b 5.54%
  56. 56 Nous-Hermes-2-Llama-2-70B 5.54%
  57. 57 Llama-2-70b-hf 5.53%
  58. 58 Yi-34B 5.50%
  59. 59 neural-chat-7b-v3-1 5.49%
  60. 60 blossom-v4-mistral-7b 5.48%
  61. 61 dolphin-2.7-mixtral-8x7b 5.46%
  62. 62 orca_mini_v3_7b 5.45%
  63. 63 deepseek-llm-7b-base 5.43%
  64. 64 dolphin-2.6-mistral-7b-dpo 5.43%
  65. 65 Mistral-7B-OpenOrca 5.41%
  66. 66 dolphin-2.6-mistral-7b 5.40%
  67. 67 Tulu 2 DPO 70B 5.40%
  68. 68 dolphin-2.6-mistral-7b-dpo-las[..] 5.39%
  69. 69 Llama-2-13b-hf 5.38%
  70. 70 deepseek-llm-7b-chat 5.32%
  71. 71 Qwen1.5-14B-Chat 5.32%
  72. 72 Orca-2-7b 5.25%
  73. 73 vicuna-13b-v1.5 5.25%
  74. 74 mistral-7b-sft-alpha 5.23%
  75. 75 Starling-LM-7B-alpha 5.19%
  76. 76 Mixtral 8x7B 5.16%
  77. 77 OpenHermes-Mixtral-8x7B 5.15%
  78. 78 Mistral-7B-AEZAKMI-v1 5.14%
  79. 79 Qwen1.5-7B 5.07%
  80. 80 samantha-mistral-instruct-7b 5.06%
  81. 81 Gemma 2B 5.06%
  82. 82 Nous-Capybara-7B-V1.9 5.05%
  83. 83 OpenHermes-7B 5.04%
  84. 84 tulu-2-7b 5.03%
  85. 85 Qwen1.5-4B-Chat 5.03%
  86. 86 Nous-Capybara-7B-V1 5.00%
  87. 87 OpenHermes-2-Mistral-7B 5.00%
  88. 88 WizardLM-70B-V1.0 4.99%
  89. 89 Qwen1.5-14B 4.98%
  90. 90 miqu-1-70b-sf 4.95%
  91. 91 Mistral-7B-Instruct-v0.1 4.90%
  92. 92 neural-chat-7b-v3-3 4.77%
  93. 93 WizardLM-13B-V1.2 4.74%
  94. 94 Zephyr 7B Alpha 4.66%
  95. 95 Toppy-M-7B 4.65%
  96. 96 mythalion-13b 4.64%
  97. 97 Xwin-LM-70B-V0.1 4.64%
  98. 98 Llama-2-70b-chat-hf scenario instruction 4.62%
  99. 99 Nous-Hermes-Llama2-13b 4.57%
  100. 100 neural-chat-7b-v3-3-Slerp 4.54%
  101. 101 Platypus2-70B-instruct 4.53%
  102. 102 ReMM-v2.2-L2-13B 4.48%
  103. 103 Llama-2-7B-32K-Instruct 4.47%
  104. 104 mixtral-megamerge-dare-8x7b-v2 4.46%
  105. 105 Llama-2-13b-chat-hf scenario instruction 4.44%
  106. 106 ReMM-SLERP-L2-13B 4.44%
  107. 107 firefly-mixtral-8x7b 4.43%
  108. 108 MythoMax-L2-13b 4.43%
  109. 109 mistral-7b-sft-beta 4.40%
  110. 110 tulu-2-dpo-13b 4.40%
  111. 111 Qwen1.5-7B-Chat 4.37%
  112. 112 NeuralMonarch-7B 4.36%
  113. 113 Llama-2-7b-chat-hf 4.33%
  114. 114 Nous-Hermes-llama-2-7b 4.33%
  115. 115 gemma-2b-it 4.31%
  116. 116 samantha-1.2-mistral-7b 4.30%
  117. 117 pygmalion-2-7b 4.25%
  118. 118 AlphaMonarch-7B 4.24%
  119. 119 yi-34B-v4 4.24%
  120. 120 Zephyr 7B Beta 4.24%
  121. 121 pygmalion-2-13b 4.21%
  122. 122 Qwen1.5-1.8B-Chat 4.04%
  123. 123 Qwen1.5-1.8B 4.02%
  124. 124 Qwen1.5-0.5B-Chat 3.95%
  125. 125 Mixtral-8x7B-Instruct-v0.1 3.94%
  126. 126 tulu-2-dpo-7b 3.78%
  127. 127 Mistral-7B-Instruct-v0.2 3.76%
  128. 128 notus-7b-v1 3.75%
  129. 129 Xwin-LM-7B-V0.2 3.75%
  130. 130 gemma-7b-it 3.65%
  131. 131 Xwin-LM-13B-V0.2 3.60%
  132. 132 Qwen1.5-4B 3.59%
  133. 133 Qwen1.5-0.5B 3.52%
  134. 134 Yi-34B-Chat 3.11%

Source: The RuLES paper

Source: The RuLES paper

#EntryScoreSettingDateSource
1 GPT-4 Turbo Nov 2023 9.54% scenario instruction Nov 2023 Official leaderboard
2 GPT-4 Turbo (Jan 2024) 9.53% scenario instruction Jan 2024 Official leaderboard
3 Claude 3 Opus 8.51% Feb 2024 Official leaderboard
4 GPT-3.5 Turbo Nov 2023 8.30% scenario instruction Nov 2023 Official leaderboard
5 GPT-3.5 Turbo Jan 2024 7.90% scenario instruction Jan 2024 Official leaderboard
6 claude-instant-v1.2 7.88% Official leaderboard
7 Qwen1.5-72B 7.02% Feb 2024 Official leaderboard
8 Claude 2.1 6.87% Nov 2023 Official leaderboard
9 Yi-34B-200K-AEZAKMI-v2 6.86% Official leaderboard
10 GPT-3.5 Turbo Jun 2023 6.80% Jun 2023 Official leaderboard
11 Claude 2 6.79% Jul 2023 Official leaderboard
12 tulu-2-70b 6.76% Official leaderboard
13 dolphin-2.2-70b 6.64% Official leaderboard
14 Gemini 1.0 Pro 6.51% Dec 2023 Official leaderboard
15 Mixtral-SlimOrca-8x7B 6.50% Official leaderboard
16 Instruct_Mixtral-8x7B-v0.1_Dol[..] 6.49% Official leaderboard
17 Nous-Hermes-2-Yi-34B 6.40% Official leaderboard
18 DeepSeek LLM 67B 6.39% Nov 2023 Official leaderboard
19 Claude 3 Sonnet 6.39% scenario instruction Feb 2024 Official leaderboard
20 dolphin-2.2-yi-34b-200k 6.38% Official leaderboard
21 Yi-34B-AEZAKMI-v1 6.38% Official leaderboard
22 yi-34B-v2 6.30% Official leaderboard
23 Nous-Hermes-2-Mixtral-8x7B-SFT 6.25% Official leaderboard
24 yi-34B-v3 6.24% Official leaderboard
25 OrionStar-Yi-34B-Chat 6.15% Official leaderboard
26 blossom-v4-yi-34b 6.13% Official leaderboard
27 deepseek-llm-67b-base 6.12% Official leaderboard
28 Llama-2-7b-hf 6.11% Official leaderboard
29 Yi-34B-200K-DARE-merge-v5 6.08% Official leaderboard
30 Capybara-Tess-Yi-34B-200K 6.06% Official leaderboard
31 platypus-yi-34b 6.03% Official leaderboard
32 openchat-3.5-0106 5.99% Official leaderboard
33 notux-8x7b-v1 5.96% Official leaderboard
34 orca_mini_v3_13b 5.96% Official leaderboard
35 SOLAR-0-70b-16bit 5.94% Official leaderboard
36 Mistral-7B-AEZAKMI-v2 5.92% Official leaderboard
37 samantha-mistral-7b 5.87% Official leaderboard
38 Gemma 7B 5.79% Feb 2024 Official leaderboard
39 GPT-3.5 Turbo (Mar 2023) 5.78% Mar 2023 Official leaderboard
40 vicuna-7b-v1.5 5.76% Official leaderboard
41 openchat-3.5-1210 5.74% Official leaderboard
42 dolphin-2.6-mixtral-8x7b 5.73% Official leaderboard
43 neural-chat-7b-v3-2 5.73% Official leaderboard
44 blossom-v3-mistral-7b 5.69% Official leaderboard
45 tulu-2-13b 5.69% Official leaderboard
46 dolphin-2.2.1-mistral-7b 5.66% Official leaderboard
47 CaPlatTessDolXaBoros-Yi-34B-20[..] 5.64% Official leaderboard
48 orca_mini_v3_70b 5.63% Official leaderboard
49 Mistral 7B v0.1 5.63% Sep 2023 Official leaderboard
50 dolphin-2.5-mixtral-8x7b 5.62% Official leaderboard
51 OpenHermes-2.5-Mistral-7B 5.62% Official leaderboard
52 Nous-Hermes-2-Mixtral-8x7B-DPO 5.62% Jan 2024 Official leaderboard
53 Samantha-1.11-70b 5.56% Official leaderboard
54 Orca-2-13b 5.55% Official leaderboard
55 blossom-v3_1-mistral-7b 5.54% Official leaderboard
56 Nous-Hermes-2-Llama-2-70B 5.54% Official leaderboard
57 Llama-2-70b-hf 5.53% Official leaderboard
58 Yi-34B 5.50% Nov 2023 Official leaderboard
59 neural-chat-7b-v3-1 5.49% Official leaderboard
60 blossom-v4-mistral-7b 5.48% Official leaderboard
61 dolphin-2.7-mixtral-8x7b 5.46% Official leaderboard
62 orca_mini_v3_7b 5.45% Official leaderboard
63 deepseek-llm-7b-base 5.43% Official leaderboard
64 dolphin-2.6-mistral-7b-dpo 5.43% Official leaderboard
65 Mistral-7B-OpenOrca 5.41% Official leaderboard
66 dolphin-2.6-mistral-7b 5.40% Official leaderboard
67 Tulu 2 DPO 70B 5.40% Nov 2023 Official leaderboard
68 dolphin-2.6-mistral-7b-dpo-las[..] 5.39% Official leaderboard
69 Llama-2-13b-hf 5.38% Official leaderboard
70 deepseek-llm-7b-chat 5.32% Official leaderboard
71 Qwen1.5-14B-Chat 5.32% Feb 2024 Official leaderboard
72 Orca-2-7b 5.25% Official leaderboard
73 vicuna-13b-v1.5 5.25% Official leaderboard
74 mistral-7b-sft-alpha 5.23% Official leaderboard
75 Starling-LM-7B-alpha 5.19% Nov 2023 Official leaderboard
76 Mixtral 8x7B 5.16% Dec 2023 Official leaderboard
77 OpenHermes-Mixtral-8x7B 5.15% Official leaderboard
78 Mistral-7B-AEZAKMI-v1 5.14% Official leaderboard
79 Qwen1.5-7B 5.07% Feb 2024 Official leaderboard
80 samantha-mistral-instruct-7b 5.06% Official leaderboard
81 Gemma 2B 5.06% Feb 2024 Official leaderboard
82 Nous-Capybara-7B-V1.9 5.05% Official leaderboard
83 OpenHermes-7B 5.04% Official leaderboard
84 tulu-2-7b 5.03% Official leaderboard
85 Qwen1.5-4B-Chat 5.03% Feb 2024 Official leaderboard
86 Nous-Capybara-7B-V1 5.00% Official leaderboard
87 OpenHermes-2-Mistral-7B 5.00% Official leaderboard
88 WizardLM-70B-V1.0 4.99% Official leaderboard
89 Qwen1.5-14B 4.98% Feb 2024 Official leaderboard
90 miqu-1-70b-sf 4.95% Official leaderboard
91 Mistral-7B-Instruct-v0.1 4.90% Sep 2023 Official leaderboard
92 neural-chat-7b-v3-3 4.77% Official leaderboard
93 WizardLM-13B-V1.2 4.74% Official leaderboard
94 Zephyr 7B Alpha 4.66% Oct 2023 Official leaderboard
95 Toppy-M-7B 4.65% Official leaderboard
96 mythalion-13b 4.64% Official leaderboard
97 Xwin-LM-70B-V0.1 4.64% Official leaderboard
98 Llama-2-70b-chat-hf 4.62% scenario instruction Official leaderboard
99 Nous-Hermes-Llama2-13b 4.57% Official leaderboard
100 neural-chat-7b-v3-3-Slerp 4.54% Official leaderboard
101 Platypus2-70B-instruct 4.53% Official leaderboard
102 ReMM-v2.2-L2-13B 4.48% Official leaderboard
103 Llama-2-7B-32K-Instruct 4.47% Official leaderboard
104 mixtral-megamerge-dare-8x7b-v2 4.46% Official leaderboard
105 Llama-2-13b-chat-hf 4.44% scenario instruction Official leaderboard
106 ReMM-SLERP-L2-13B 4.44% Official leaderboard
107 firefly-mixtral-8x7b 4.43% Official leaderboard
108 MythoMax-L2-13b 4.43% Official leaderboard
109 mistral-7b-sft-beta 4.40% Official leaderboard
110 tulu-2-dpo-13b 4.40% Official leaderboard
111 Qwen1.5-7B-Chat 4.37% Feb 2024 Official leaderboard
112 NeuralMonarch-7B 4.36% Official leaderboard
113 Llama-2-7b-chat-hf 4.33% Official leaderboard
114 Nous-Hermes-llama-2-7b 4.33% Official leaderboard
115 gemma-2b-it 4.31% Official leaderboard
116 samantha-1.2-mistral-7b 4.30% Official leaderboard
117 pygmalion-2-7b 4.25% Official leaderboard
118 AlphaMonarch-7B 4.24% Official leaderboard
119 yi-34B-v4 4.24% Official leaderboard
120 Zephyr 7B Beta 4.24% Oct 2023 Official leaderboard
121 pygmalion-2-13b 4.21% Official leaderboard
122 Qwen1.5-1.8B-Chat 4.04% Official leaderboard
123 Qwen1.5-1.8B 4.02% Official leaderboard
124 Qwen1.5-0.5B-Chat 3.95% Official leaderboard
125 Mixtral-8x7B-Instruct-v0.1 3.94% Official leaderboard
126 tulu-2-dpo-7b 3.78% Official leaderboard
127 Mistral-7B-Instruct-v0.2 3.76% Dec 2023 Official leaderboard
128 notus-7b-v1 3.75% Official leaderboard
129 Xwin-LM-7B-V0.2 3.75% Official leaderboard
130 gemma-7b-it 3.65% Official leaderboard
131 Xwin-LM-13B-V0.2 3.60% Official leaderboard
132 Qwen1.5-4B 3.59% Official leaderboard
133 Qwen1.5-0.5B 3.52% Official leaderboard
134 Yi-34B-Chat 3.11% Official leaderboard
#ModelScoreSettingDateSource
1 GPT-4 Turbo Nov 2023 9.54% scenario instruction Nov 2023 Official leaderboard
2 GPT-4 Turbo (Jan 2024) 9.53% scenario instruction Jan 2024 Official leaderboard
3 Claude 3 Opus 8.51% Feb 2024 Official leaderboard
4 GPT-3.5 Turbo Nov 2023 8.30% scenario instruction Nov 2023 Official leaderboard
5 GPT-3.5 Turbo Jan 2024 7.90% scenario instruction Jan 2024 Official leaderboard
6 Qwen1.5-72B 7.02% Feb 2024 Official leaderboard
7 Claude 2.1 6.87% Nov 2023 Official leaderboard
8 GPT-3.5 Turbo Jun 2023 6.80% Jun 2023 Official leaderboard
9 Claude 2 6.79% Jul 2023 Official leaderboard
10 Gemini 1.0 Pro 6.51% Dec 2023 Official leaderboard
11 DeepSeek LLM 67B 6.39% Nov 2023 Official leaderboard
12 Claude 3 Sonnet 6.39% scenario instruction Feb 2024 Official leaderboard
13 Gemma 7B 5.79% Feb 2024 Official leaderboard
14 GPT-3.5 Turbo (Mar 2023) 5.78% Mar 2023 Official leaderboard
15 Mistral 7B v0.1 5.63% Sep 2023 Official leaderboard
16 Nous-Hermes-2-Mixtral-8x7B-DPO 5.62% Jan 2024 Official leaderboard
17 Yi-34B 5.50% Nov 2023 Official leaderboard
18 Tulu 2 DPO 70B 5.40% Nov 2023 Official leaderboard
19 Qwen1.5-14B-Chat 5.32% Feb 2024 Official leaderboard
20 Starling-LM-7B-alpha 5.19% Nov 2023 Official leaderboard
21 Mixtral 8x7B 5.16% Dec 2023 Official leaderboard
22 Qwen1.5-7B 5.07% Feb 2024 Official leaderboard
23 Gemma 2B 5.06% Feb 2024 Official leaderboard
24 Qwen1.5-4B-Chat 5.03% Feb 2024 Official leaderboard
25 Qwen1.5-14B 4.98% Feb 2024 Official leaderboard
26 Mistral-7B-Instruct-v0.1 4.90% Sep 2023 Official leaderboard
27 Zephyr 7B Alpha 4.66% Oct 2023 Official leaderboard
28 Qwen1.5-7B-Chat 4.37% Feb 2024 Official leaderboard
29 Zephyr 7B Beta 4.24% Oct 2023 Official leaderboard
30 Mistral-7B-Instruct-v0.2 3.76% Dec 2023 Official leaderboard

Rankings by part

We read 6 parts from the benchmark’s paper, each ranking models on one part of the test, such as Benign harmless, Benign helpful and Basic harmless. GPT-4 Turbo leads 3 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.

Benign harmless

30 models we track have a score here.

  1. 1 Mistral-7B-Instruct-v0.1 ↑25 10%
  2. 2 Mistral 7B v0.1 ↑13 10%
  3. 3 Yi-34B ↑14 10%
  4. 4 GPT-4 Turbo Nov 2023 ↓3 10%
  5. 5 GPT-4 Turbo (Jan 2024) ↓3 10%
  6. 6 Gemma 7B ↑7 10%
  7. 7 Zephyr 7B Alpha ↑20 9.96%
  8. 8 Zephyr 7B Beta ↑21 9.96%
  9. 9 Mixtral 8x7B ↑12 9.96%
  10. 10 Qwen1.5-7B ↑12 9.96%

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-4 Turbo
scenario in…
9.54%
GPT-4 Turbo (Jan 2024)
scenario in…
9.53%
Claude 3 Opus
scenario in…
8.51%
GPT-3.5 Turbo
scenario in…
8.30%
GPT-3.5 Turbo
scenario in…
7.90%
Claude 2.1
scenario in…
6.87%
Claude 3 Sonnet
scenario in…
6.39%

Source: The RuLES paper

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-4 Turbo
scenario in…
9.54%
GPT-4 Turbo (Jan 2024)
scenario in…
9.53%
Claude 3 Opus
scenario in…
8.51%
GPT-3.5 Turbo
scenario in…
8.30%
GPT-3.5 Turbo
scenario in…
7.90%
Claude 2.1
scenario in…
6.87%
Claude 3 Sonnet
scenario in…
6.39%

Source: The RuLES paper

2 models have held number one in the benchmark’s paper since its release in November 2023

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

OpenAI
Benchmark paper Replaced by 2.0, Mar 2024
GPT-4 Turbo 9.54%

Source: The RuLES paper

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
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2023 2024 Near the ceiling (85%) GPT-4 Turbo GPT-3.5 Turbo GPT-3.5 Turbo (Mar 2023) 0% 20% 40% 60% 80% 100% 2023 2024 Near the ceiling (85%) GPT-4 Turbo

Source: Official RuLES 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.

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

OpenAI
RuLES 2.0 Not tested since Mar 2024
GPT-4 Turbo 9.54%
RuLES Replaced by 2.0, Mar 2024
GPT-4 Turbo 9.54%
RuLES 2.0 Not tested since Mar 2024
GPT-4 Turbo 9.54%

Source: Official leaderboards

More in Instructions and chat

All 144 in Instructions and chat
  • IFEval In use Models answer about 500 prompts containing verifiable instructions such as word limits or required keywords. Code checks each instruction for strict and loose accuracy. Gemma 4 31B IT 98.9%
  • IFBench In use 58 new out-of-domain verifiable output constraints. Scored by programmatic constraint checks (strict or loose). Flint 89.2%
  • MultiChallenge In use Multi-turn conversations testing instruction retention, inference memory, versioned editing and self-coherence. Graded by LLM judge with rubrics. GPT-6.1 Sol 82%
  • Multi-IF In use Models follow verifiable instructions over three-turn conversations in eight languages, extending IFEval. Scored by instruction accuracy at each turn. LFM3-1.7B 89.1%
  • AdvancedIF In use Models respond to over 1,600 prompts with complex, multi-turn, and system-level instructions. A judge model checks each response against expert-written rubrics. MAI-Thinking-1 85%
  • MM-IFEval In use Multimodal models follow image-grounded instructions with exact output constraints. Checks mix rule-based verification and a judge model. GPT-4o 64.6%

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 benchmark’s paper (134 entries).
  • Pass Models scored 30 models we track, plus 104 other entries on its leaderboard.
  • Note Newest score The newest came 2 years ago, in February 2024, for Claude 3 Opus. That’s expected for a test replaced by RuLES 2.0.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Claude 3 Opus (#173).
  • Pass Headroom The best is 9.54%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 38 rows for models we track and 104 other entries at the last read.
  • Pass Last read November 21st, 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