Benchmark Almanac

DesignArena Python-PPTX Slides

Claude Opus 5.5 1,427 leads since September 2026
Not in model cards

Agents build PowerPoint decks with python-pptx from a user prompt. Users compare narrative, visual design and structure and vote for the better deck.

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

Kind
Documents and spreadsheets
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
November 2026
Scores refreshed
November 12th, 2026
36
Models scored on its own leaderboard

Claude Opus 5.5 leads with 1,427

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

36 models. Scroll sideways for the rest. Bars start at 900.

  1. 1,427 Claude Opus 5.5
  2. 1,382 GPT-6 Astra
  3. 1,345 GPT-6.1 Sol
  4. 1,308 Claude Fable 5.1
  5. 1,290 Kimi K3.1 Pro
  6. 1,252 Claude Fable 5
  7. 1,221 Kimi K3
  8. 1,220 Qwen 3.8 Max
  9. 1,217 Claude Opus 5
  10. 1,210 Inkling 2 Small
  11. 1,205 GLM-5.3
  12. 1,204 Muse Spark 1.3
  13. 1,198 GPT-5.6 Sol
  14. 1,197 Grok 4.6
  15. 1,195 Grok 4.5
  16. 1,185 Claude Opus 4.8
  17. 1,171 MiniMax-M3
  18. 1,171 Claude Sonnet 5
  19. 1,159 GLM-5.2
  20. 1,154 Muse Spark 1.2
  21. 1,150 GLM-5.1
  22. 1,147 Qwen3.7-Max
  23. 1,137 Gemini 3.5 Flash
  24. 1,132 Gemini 3.8 Flash
  25. 1,117 Muse Spark 1.1
  26. 1,112 Kimi K2.7 Code
  27. 1,109 GLM-5V-Turbo
  28. 1,092 Gemini 3.6 Flash
  29. 1,088 GPT-5.6 Terra
  30. 1,087 Kimi K2.6
  31. 1,068 GPT-5.5
  32. 1,065 Gemini 3.7 Flash
  33. 1,065 Muse Wonder
  34. 1,027 Gemini 3.1 Pro
  35. 1,003 Gemini 3 Pro
  36. 970 Grok 4.3

Bars start at 900.

  1. 1 Claude Opus 5.5 high 1,427
  2. 2 GPT-6 Astra xhigh, max 1,382
  3. 3 GPT-6.1 Sol max 1,345
  4. 4 Claude Fable 5.1 1,308
  5. 5 Kimi K3.1 Pro 1,290
  6. 6 Claude Fable 5 1,252
  7. 7 Kimi K3 1,221
  8. 8 Qwen 3.8 Max 1,220
  9. 9 Claude Opus 5 1,217
  10. 10 Inkling 2 Small 1,210
  11. 11 GLM-5.3 1,205
  12. 12 Muse Spark 1.3 xhigh 1,204
  13. 13 GPT-5.6 Sol medium 1,198
  14. 14 Grok 4.6 1,197
  15. 15 Grok 4.5 1,195
  16. 16 Claude Opus 4.8 1,185
  17. 17 MiniMax-M3 1,171
  18. 18 Claude Sonnet 5 1,171
  19. 19 GLM-5.2 1,159
  20. 20 Muse Spark 1.2 1,154
  21. 21 GLM-5.1 1,150
  22. 22 Qwen3.7-Max 1,147
  23. 23 Gemini 3.5 Flash 1,137
  24. 24 Gemini 3.8 Flash 1,132
  25. 25 Muse Spark 1.1 1,117
  26. 26 Kimi K2.7 Code 1,112
  27. 27 GLM-5V-Turbo 1,109
  28. 28 Gemini 3.6 Flash 1,092
  29. 29 GPT-5.6 Terra 1,088
  30. 30 Kimi K2.6 1,087
  31. 31 GPT-5.5 1,068
  32. 32 Gemini 3.7 Flash 1,065
  33. 33 Muse Wonder max 1,065
  34. 34 Gemini 3.1 Pro 1,027
  35. 35 Gemini 3 Pro 1,003
  36. 36 Grok 4.3 970

Source: Official DesignArena Python-PPTX Slides leaderboard

#ModelScoreSettingDateSource
1 Claude Opus 5.5 1,427 high Sep 2026 Official leaderboard
2 GPT-6 Astra 1,382 xhigh, max Sep 2026 Official leaderboard
3 GPT-6.1 Sol 1,345 max Sep 2026 Official leaderboard
4 Claude Fable 5.1 1,308 Sep 2026 Official leaderboard
5 Kimi K3.1 Pro 1,290 Oct 2026 Official leaderboard
6 Claude Fable 5 1,252 Jun 2026 Official leaderboard
7 Kimi K3 1,221 Jul 2026 Official leaderboard
8 Qwen 3.8 Max 1,220 Aug 2026 Official leaderboard
9 Claude Opus 5 1,217 Jul 2026 Official leaderboard
10 Inkling 2 Small 1,210 Nov 2026 Official leaderboard
11 GLM-5.3 1,205 Aug 2026 Official leaderboard
12 Muse Spark 1.3 1,204 xhigh Sep 2026 Official leaderboard
13 GPT-5.6 Sol 1,198 medium Jul 2026 Official leaderboard
14 Grok 4.6 1,197 Aug 2026 Official leaderboard
15 Grok 4.5 1,195 Jul 2026 Official leaderboard
16 Claude Opus 4.8 1,185 May 2026 Official leaderboard
17 MiniMax-M3 1,171 Jun 2026 Official leaderboard
18 Claude Sonnet 5 1,171 Jun 2026 Official leaderboard
19 GLM-5.2 1,159 Jun 2026 Official leaderboard
20 Muse Spark 1.2 1,154 Aug 2026 Official leaderboard
21 GLM-5.1 1,150 Apr 2026 Official leaderboard
22 Qwen3.7-Max 1,147 May 2026 Official leaderboard
23 Gemini 3.5 Flash 1,137 May 2026 Official leaderboard
24 Gemini 3.8 Flash 1,132 Sep 2026 Official leaderboard
25 Muse Spark 1.1 1,117 Jul 2026 Official leaderboard
26 Kimi K2.7 Code 1,112 Jun 2026 Official leaderboard
27 GLM-5V-Turbo 1,109 Apr 2026 Official leaderboard
28 Gemini 3.6 Flash 1,092 Jul 2026 Official leaderboard
29 GPT-5.6 Terra 1,088 Jul 2026 Official leaderboard
30 Kimi K2.6 1,087 Apr 2026 Official leaderboard
31 GPT-5.5 1,068 Apr 2026 Official leaderboard
32 Gemini 3.7 Flash 1,065 Aug 2026 Official leaderboard
33 Muse Wonder 1,065 max Oct 2026 Official leaderboard
34 Gemini 3.1 Pro 1,027 Feb 2026 Official leaderboard
35 Gemini 3 Pro 1,003 Nov 2025 Official leaderboard
36 Grok 4.3 970 Apr 2026 Official leaderboard

9 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.

Google DeepMind Z.ai Anthropic OpenAI
Own leaderboard 9 leaders
Claude Opus 5.5 1,427

Source: Official DesignArena Python-PPTX Slides 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
1,000 1,200 1,400 2026 2027 1,000 1,200 1,400 2026 2027

Source: Official DesignArena Python-PPTX Slides 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 Economic work

All 97 in Economic work
  • JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
  • GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. Seed2.1 Pro 87.9%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
  • CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%
  • OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%