Benchmark Almanac

DesignArena ASCII Art

Claude Opus 5.5 1,354 leads since September 2026
Not in model cards

Models draw a requested picture using only text characters. Users compare the ASCII art and vote for the better one.

One of 40 DesignArena boards. DesignArena ranks entries with Bradley-Terry ratings from user votes.

Kind
Creative writing
Level
Everyday
Scored
Ranked by votes
Answers
Open-ended
Tasks
All public
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
84
Entries on its own leaderboard, 77 models we track

Claude Opus 5.5 leads with 1,354

Best score per entry from the benchmark's own leaderboard, read on November 12th, 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.

84 entries. Scroll sideways for the rest. Bars start at 900.

  1. 1,354 Claude Opus 5.5
  2. 1,353 Claude Opus 5
  3. 1,344 GPT-6 Astra
  4. 1,329 Muse Spark 1.3
  5. 1,321 Claude Fable 5
  6. 1,318 Claude Fable 5.1
  7. 1,305 GPT-5.6 Sol
  8. 1,302 Kimi K3.1 Pro
  9. 1,278 Muse Spark 1.2
  10. 1,276 Gemini 3.6 Flash
  11. 1,273 Gemini 3.8 Flash
  12. 1,271 Claude Opus 4.8
  13. 1,269 GLM-5.3-Flash
  14. 1,268 Gemini 3.7 Flash
  15. 1,267 Grok 4.6
  16. 1,265 Gemini 3.1 Pro
  17. 1,260 DeepSeek V4.1 Flash
  18. 1,256 Muse Spark 1.1
  19. 1,250 Claude Opus 4.6
  20. 1,249 Grok 4.5
  21. 1,248 Gemini 3.5 Flash
  22. 1,247 GPT-5.5
  23. 1,230 Claude Sonnet 4.6
  24. 1,220 Deca 2.5 Ultra
  25. 1,217 GPT-6 Sol
  26. 1,215 Claude Haiku 5.5
  27. 1,210 GLM-5.3
  28. 1,204 Qwen3.7-Max
  29. 1,204 Mistral Large 4
  30. 1,200 Claude Sonnet 4.5
  31. 1,199 Claude Sonnet 5
  32. 1,194 GPT-5.4
  33. 1,192 Qwen 3.8 Max
  34. 1,186 GLM-5.2
  35. 1,185 Claude Opus 4.5
  36. 1,185 Kimi K2.7 Code
  37. 1,171 GPT-5.6 Terra
  38. 1,170 Grok 4.20 Beta
  39. 1,162 Claude Opus 4.1
  40. 1,162 Gemini 3.1 Flash-Lite
  41. 1,161 GPT-6 Luna
  42. 1,160 GLM-4.7
  43. 1,160 Kimi K2.5
  44. 1,158 GLM-5-Turbo
  45. 1,153 Tencent Hy3
  46. 1,152 Kimi K2.6
  47. 1,151 MiMo-V2.5-Pro
  48. 1,149 MiniMax-M3
  49. 1,147 GPT-5.2
  50. 1,147 GPT-5.3 Codex
  51. 1,143 DeepSeek-V4-Pro
  52. 1,142 GLM-5
  53. 1,142 Step 3.7 Flash
  54. 1,142 GPT-5.6 Luna
  55. 1,141 GLM-5.1
  56. 1,139 MiMo-V2.5
  57. 1,136 GPT-5
  58. 1,135 Claude Haiku 4.5
  59. 1,134 Qwen3.7-Plus
  60. 1,129 MiniMax-M2.7
  61. 1,122 Qwen3-Max
  62. 1,121 Grok 4.3
  63. 1,119 GPT-5 mini (Default)
  64. 1,114 GPT-5.1
  65. 1,113 GLM-5V-Turbo
  66. 1,110 Qwen 3.6 Plus
  67. 1,104 DeepSeek-V4-Flash
  68. 1,103 Grok 4 Fast
  69. 1,102 Grok 4.1 Fast
  70. 1,100 GPT-5.1-Codex-mini
  71. 1,095 Nex N2 Pro
  72. 1,093 Nemotron 3 Ultra
  73. 1,093 DeepSeek V4 Flash 0731
  74. 1,089 Qwen3.5 Plus 02-15
  75. 1,087 Inkling
  76. 1,077 DeepSeek-V3.2
  77. 1,064 Mistral Large 3
  78. 1,057 Mistral Medium 3.5
  79. 1,036 Trinity Large Thinking
  80. 1,033 AesCoder-4B
  81. 1,012 MiMo-V2-Flash
  82. 992 Mistral Medium 3.1
  83. 981 Mercury 2
  84. 955 Prime Intellect: INTELLECT-3

Bars start at 900.

  1. 1 Claude Opus 5.5 1,354
  2. 2 Claude Opus 5 1,353
  3. 3 GPT-6 Astra xhigh 1,344
  4. 4 Muse Spark 1.3 xhigh, max 1,329
  5. 5 Claude Fable 5 1,321
  6. 6 Claude Fable 5.1 1,318
  7. 7 GPT-5.6 Sol medium, xhigh 1,305
  8. 8 Kimi K3.1 Pro 1,302
  9. 9 Muse Spark 1.2 1,278
  10. 10 Gemini 3.6 Flash 1,276
  11. 11 Gemini 3.8 Flash 1,273
  12. 12 Claude Opus 4.8 1,271
  13. 13 GLM-5.3-Flash 1,269
  14. 14 Gemini 3.7 Flash 1,268
  15. 15 Grok 4.6 1,267
  16. 16 Gemini 3.1 Pro 1,265
  17. 17 DeepSeek V4.1 Flash 1,260
  18. 18 Muse Spark 1.1 1,256
  19. 19 Claude Opus 4.6 thinking 1,250
  20. 20 Grok 4.5 1,249
  21. 21 Gemini 3.5 Flash 1,248
  22. 22 GPT-5.5 1,247
  23. 23 Claude Sonnet 4.6 1,230
  24. 24 Deca 2.5 Ultra 2 1,220
  25. 25 GPT-6 Sol 1,217
  26. 26 Claude Haiku 5.5 1,215
  27. 27 GLM-5.3 1,210
  28. 28 Qwen3.7-Max 1,204
  29. 29 Mistral Large 4 1,204
  30. 30 Claude Sonnet 4.5 thinking 1,200
  31. 31 Claude Sonnet 5 1,199
  32. 32 GPT-5.4 medium 1,194
  33. 33 Qwen 3.8 Max 1,192
  34. 34 GLM-5.2 1,186
  35. 35 Claude Opus 4.5 1,185
  36. 36 Kimi K2.7 Code 1,185
  37. 37 GPT-5.6 Terra 1,171
  38. 38 Grok 4.20 Beta reasoning 1,170
  39. 39 Claude Opus 4.1 1,162
  40. 40 Gemini 3.1 Flash-Lite 1,162
  41. 41 GPT-6 Luna 1,161
  42. 42 GLM-4.7 1,160
  43. 43 Kimi K2.5 thinking 1,160
  44. 44 GLM-5-Turbo 1,158
  45. 45 Tencent Hy3 1,153
  46. 46 Kimi K2.6 1,152
  47. 47 MiMo-V2.5-Pro 1,151
  48. 48 MiniMax-M3 1,149
  49. 49 GPT-5.2 low 1,147
  50. 50 GPT-5.3 Codex 1,147
  51. 51 DeepSeek-V4-Pro 1,143
  52. 52 GLM-5 1,142
  53. 53 Step 3.7 Flash 1,142
  54. 54 GPT-5.6 Luna 1,142
  55. 55 GLM-5.1 1,141
  56. 56 MiMo-V2.5 1,139
  57. 57 GPT-5 minimal 1,136
  58. 58 Claude Haiku 4.5 1,135
  59. 59 Qwen3.7-Plus 1,134
  60. 60 MiniMax-M2.7 1,129
  61. 61 Qwen3-Max 1,122
  62. 62 Grok 4.3 1,121
  63. 63 GPT-5 mini (Default) 1,119
  64. 64 GPT-5.1 none 1,114
  65. 65 GLM-5V-Turbo 1,113
  66. 66 Qwen 3.6 Plus 1,110
  67. 67 DeepSeek-V4-Flash 1,104
  68. 68 Grok 4 Fast reasoning 1,103
  69. 69 Grok 4.1 Fast 1,102
  70. 70 GPT-5.1-Codex-mini 1,100
  71. 71 Nex N2 Pro 1,095
  72. 72 Nemotron 3 Ultra 1,093
  73. 73 DeepSeek V4 Flash 0731 1,093
  74. 74 Qwen3.5 Plus 02-15 1,089
  75. 75 Inkling 1,087
  76. 76 DeepSeek-V3.2 1,077
  77. 77 Mistral Large 3 1,064
  78. 78 Mistral Medium 3.5 1,057
  79. 79 Trinity Large Thinking 1,036
  80. 80 AesCoder-4B 1,033
  81. 81 MiMo-V2-Flash 1,012
  82. 82 Mistral Medium 3.1 992
  83. 83 Mercury 2 981
  84. 84 Prime Intellect: INTELLECT-3 955

Source: Official DesignArena ASCII Art leaderboard

#EntryScoreSettingDateSource
1 Claude Opus 5.5 1,354 Sep 2026 Official leaderboard
2 Claude Opus 5 1,353 Jul 2026 Official leaderboard
3 GPT-6 Astra 1,344 xhigh Sep 2026 Official leaderboard
4 Muse Spark 1.3 1,329 xhigh, max Sep 2026 Official leaderboard
5 Claude Fable 5 1,321 Jun 2026 Official leaderboard
6 Claude Fable 5.1 1,318 Sep 2026 Official leaderboard
7 GPT-5.6 Sol 1,305 medium, xhigh Jul 2026 Official leaderboard
8 Kimi K3.1 Pro 1,302 Oct 2026 Official leaderboard
9 Muse Spark 1.2 1,278 Aug 2026 Official leaderboard
10 Gemini 3.6 Flash 1,276 Jul 2026 Official leaderboard
11 Gemini 3.8 Flash 1,273 Sep 2026 Official leaderboard
12 Claude Opus 4.8 1,271 May 2026 Official leaderboard
13 GLM-5.3-Flash 1,269 Aug 2026 Official leaderboard
14 Gemini 3.7 Flash 1,268 Aug 2026 Official leaderboard
15 Grok 4.6 1,267 Aug 2026 Official leaderboard
16 Gemini 3.1 Pro 1,265 Feb 2026 Official leaderboard
17 DeepSeek V4.1 Flash 1,260 Sep 2026 Official leaderboard
18 Muse Spark 1.1 1,256 Jul 2026 Official leaderboard
19 Claude Opus 4.6 1,250 thinking Feb 2026 Official leaderboard
20 Grok 4.5 1,249 Jul 2026 Official leaderboard
21 Gemini 3.5 Flash 1,248 May 2026 Official leaderboard
22 GPT-5.5 1,247 Apr 2026 Official leaderboard
23 Claude Sonnet 4.6 1,230 Feb 2026 Official leaderboard
24 Deca 2.5 Ultra 1,220 2 Official leaderboard
25 GPT-6 Sol 1,217 Sep 2026 Official leaderboard
26 Claude Haiku 5.5 1,215 Oct 2026 Official leaderboard
27 GLM-5.3 1,210 Aug 2026 Official leaderboard
28 Qwen3.7-Max 1,204 May 2026 Official leaderboard
29 Mistral Large 4 1,204 Oct 2026 Official leaderboard
30 Claude Sonnet 4.5 1,200 thinking Sep 2025 Official leaderboard
31 Claude Sonnet 5 1,199 Jun 2026 Official leaderboard
32 GPT-5.4 1,194 medium Mar 2026 Official leaderboard
33 Qwen 3.8 Max 1,192 Aug 2026 Official leaderboard
34 GLM-5.2 1,186 Jun 2026 Official leaderboard
35 Claude Opus 4.5 1,185 Nov 2025 Official leaderboard
36 Kimi K2.7 Code 1,185 Jun 2026 Official leaderboard
37 GPT-5.6 Terra 1,171 Jul 2026 Official leaderboard
38 Grok 4.20 Beta 1,170 reasoning Official leaderboard
39 Claude Opus 4.1 1,162 Aug 2025 Official leaderboard
40 Gemini 3.1 Flash-Lite 1,162 Mar 2026 Official leaderboard
41 GPT-6 Luna 1,161 Sep 2026 Official leaderboard
42 GLM-4.7 1,160 Dec 2025 Official leaderboard
43 Kimi K2.5 1,160 thinking Jan 2026 Official leaderboard
44 GLM-5-Turbo 1,158 Mar 2026 Official leaderboard
45 Tencent Hy3 1,153 Jul 2026 Official leaderboard
46 Kimi K2.6 1,152 Apr 2026 Official leaderboard
47 MiMo-V2.5-Pro 1,151 Apr 2026 Official leaderboard
48 MiniMax-M3 1,149 Jun 2026 Official leaderboard
49 GPT-5.2 1,147 low Dec 2025 Official leaderboard
50 GPT-5.3 Codex 1,147 Feb 2026 Official leaderboard
51 DeepSeek-V4-Pro 1,143 Apr 2026 Official leaderboard
52 GLM-5 1,142 Feb 2026 Official leaderboard
53 Step 3.7 Flash 1,142 May 2026 Official leaderboard
54 GPT-5.6 Luna 1,142 Jul 2026 Official leaderboard
55 GLM-5.1 1,141 Apr 2026 Official leaderboard
56 MiMo-V2.5 1,139 Apr 2026 Official leaderboard
57 GPT-5 1,136 minimal Aug 2025 Official leaderboard
58 Claude Haiku 4.5 1,135 Oct 2025 Official leaderboard
59 Qwen3.7-Plus 1,134 Jun 2026 Official leaderboard
60 MiniMax-M2.7 1,129 Mar 2026 Official leaderboard
61 Qwen3-Max 1,122 Sep 2025 Official leaderboard
62 Grok 4.3 1,121 Apr 2026 Official leaderboard
63 GPT-5 mini (Default) 1,119 Official leaderboard
64 GPT-5.1 1,114 none Nov 2025 Official leaderboard
65 GLM-5V-Turbo 1,113 Apr 2026 Official leaderboard
66 Qwen 3.6 Plus 1,110 Mar 2026 Official leaderboard
67 DeepSeek-V4-Flash 1,104 Apr 2026 Official leaderboard
68 Grok 4 Fast 1,103 reasoning Sep 2025 Official leaderboard
69 Grok 4.1 Fast 1,102 Nov 2025 Official leaderboard
70 GPT-5.1-Codex-mini 1,100 Nov 2025 Official leaderboard
71 Nex N2 Pro 1,095 Official leaderboard
72 Nemotron 3 Ultra 1,093 Jun 2026 Official leaderboard
73 DeepSeek V4 Flash 0731 1,093 Jul 2026 Official leaderboard
74 Qwen3.5 Plus 02-15 1,089 Official leaderboard
75 Inkling 1,087 Jul 2026 Official leaderboard
76 DeepSeek-V3.2 1,077 Dec 2025 Official leaderboard
77 Mistral Large 3 1,064 Dec 2025 Official leaderboard
78 Mistral Medium 3.5 1,057 Apr 2026 Official leaderboard
79 Trinity Large Thinking 1,036 Apr 2026 Official leaderboard
80 AesCoder-4B 1,033 Official leaderboard
81 MiMo-V2-Flash 1,012 Dec 2025 Official leaderboard
82 Mistral Medium 3.1 992 Aug 2025 Official leaderboard
83 Mercury 2 981 Feb 2026 Official leaderboard
84 Prime Intellect: INTELLECT-3 955 Official leaderboard

More effort adds a median 15 elo

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 Astra
xhigh, max
1,344
Muse Spark 1.3
xhigh
1,329
GPT-5.6 Sol
medium medium, xhi…
1,305
DeepSeek V4.1 Flash
1017
1,260
Claude Opus 4.6
thinking
1,250
Claude Sonnet 4.5
thinking
1,200
GPT-5.4
none
1,194
GPT-5.2
none low
1,147
GPT-5.1
none high
1,114
Grok 4 Fast
reasoning
1,103
Grok 4.1 Fast
reasoning
1,102

Source: Official DesignArena ASCII Art leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 1,354
Claude Opus 5 1,353
GPT-6 Astra xhigh 1,344
GPT-6 Astra xhigh, max 1,336
Muse Spark 1.3 xhigh, max 1,329
Claude Fable 5 1,321
Claude Fable 5.1 1,318
GPT-5.6 Sol medium, xhigh 1,305
Muse Spark 1.3 xhigh 1,303
Kimi K3.1 Pro 1,302
Muse Spark 1.2 1,278
Gemini 3.6 Flash 1,276
Gemini 3.8 Flash 1,273
Claude Opus 4.8 1,271
GLM-5.3-Flash 1,269
Gemini 3.7 Flash 1,268
Grok 4.6 1,267
Gemini 3.1 Pro 1,265
DeepSeek V4.1 Flash 1,260
GPT-5.6 Sol medium 1,256
Muse Spark 1.1 1,256
Claude Opus 4.6 thinking 1,250
Grok 4.5 1,249
Claude Opus 4.6 1,248
Gemini 3.5 Flash 1,248
GPT-5.5 1,247
DeepSeek V4.1 Flash 1017 1,235
Claude Sonnet 4.6 1,230
GPT-6 Sol 1,217
Claude Haiku 5.5 1,215
GLM-5.3 1,210
Qwen3.7-Max 1,204
Mistral Large 4 1,204
Claude Sonnet 4.5 thinking 1,200
Claude Sonnet 5 1,199
GPT-5.4 medium 1,194
Qwen 3.8 Max 1,192
GLM-5.2 1,186
Claude Opus 4.5 1,185
Kimi K2.7 Code 1,185
GPT-5.4 none 1,183
GPT-5.4 low 1,182
GPT-5.6 Terra 1,171
Claude Sonnet 4.5 1,170
Claude Opus 4.1 1,162
Gemini 3.1 Flash-Lite 1,162
GPT-6 Luna 1,161
GLM-4.7 1,160
Kimi K2.5 thinking 1,160
GLM-5-Turbo 1,158
Tencent Hy3 1,153
Kimi K2.6 1,152
MiMo-V2.5-Pro 1,151
MiniMax-M3 1,149
GPT-5.2 low 1,147
GPT-5.3 Codex 1,147
DeepSeek-V4-Pro 1,143
GLM-5 1,142
Step 3.7 Flash 1,142
GPT-5.6 Luna 1,142
GLM-5.1 1,141
MiMo-V2.5 1,139
GPT-5 minimal 1,136
Claude Haiku 4.5 1,135
Qwen3.7-Plus 1,134
MiniMax-M2.7 1,129
Qwen3-Max 1,122
GPT-5.2 medium 1,122
Grok 4.3 1,121
GPT-5.1 none 1,114
GLM-5V-Turbo 1,113
GPT-5.1 low 1,111
Qwen 3.6 Plus 1,110
GPT-5.2 none 1,107
DeepSeek-V4-Flash 1,104
Grok 4 Fast reasoning 1,103
Grok 4.1 Fast 1,102
GPT-5.1-Codex-mini 1,100
GPT-5.1 medium 1,096
Nemotron 3 Ultra 1,093
DeepSeek V4 Flash 0731 1,093
Inkling 1,087
DeepSeek-V3.2 1,077
Mistral Large 3 1,064
Mistral Medium 3.5 1,057
GPT-5.1 high 1,056
Grok 4 Fast 1,052
Grok 4.1 Fast reasoning 1,050
Trinity Large Thinking 1,036
MiMo-V2-Flash 1,012
Mistral Medium 3.1 992
Mercury 2 981

8 models have held number one on the official leaderboard since 2023

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

Anthropic Google DeepMind
Own leaderboard 8 leaders
Claude Opus 5.5 1,354

Source: Official DesignArena ASCII Art leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
900 1,000 1,100 1,200 1,300 1,400 2026 2027 900 1,000 1,100 1,200 1,300 1,400 2026 2027

Source: Official DesignArena ASCII Art leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Writing and taste

All 30 in Writing and taste
  • Design Bench (Moonshot) In use Models make graphic designs from direct prompts or few-shot reference images. An automated grader rates the results as Elo scores. Kimi K3.1 Pro 957
  • WritingBench Fading Models complete 1,000 real-world writing requests across 6 domains and 100 subdomains. A judge model scores each reply on five query-specific criteria. GPT-5 83.9
  • Creative Writing (MiniMax in-house) Only in system cards Models write creative pieces for real Hailuo AI user requests. Automatic assessments score the writing. MiniMax-Text-01 81.3%
  • DeepSeek Chinese functional writing evaluation Only in system cards Models answer 3,170 everyday Chinese writing prompts such as reports, emails and speeches. Pairwise judging against a baseline model gives a win rate. DeepSeek-V4-Pro 62.6%
  • Slide generation human evaluation (Z.ai) Only in system cards People compare HTML slide decks from two models on content, layout and visual quality. Metric is overall win rate. GLM-5 67.5%
  • Task-dependent output homogenization evaluation Only in system cards Models answer prompts several times. A judge scores functional diversity where it is wanted and consistency where one answer is right. Muse Spark 0.21