Benchmark Almanac

VoxelBench

GPT-6 Astra 2,757 leads since Sep 2026 Not in model cards

Language models turn text or image prompts into voxel structures. Voters compare anonymous builds from the same prompt, and Glicko-2 ratings rank the models.

Separate text-prompt and image-prompt boards. A model needs 50 votes to appear. Different from MineBench.

Made by VoxelBench Released Sep 2025 Spatial reasoning and counting All tasks public Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Everyday Most adults, without training
Scored
Ranked by votes People pick the better of two answers, as in an arena, and the votes are turned into ratings.
Answers
Open-ended Free-form text, documents, designs or media judged as a whole.
Leaderboard voxelbench.ai
57
Entries on its own leaderboard, 56 models we track

GPT-6 Astra leads with 2,757

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 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.

57 entries. Scroll sideways for the rest. Bars start at 1,000.

  1. 2,757 GPT-6 Astra
  2. 2,647 GPT-6.1 Sol
  3. 2,605 Claude Opus 5.5
  4. 2,426 GPT-6 Sol
  5. 2,297 Claude Opus 5
  6. 2,295 GPT-6.1 Terra
  7. 2,288 GPT-5.6 Sol
  8. 2,235 Gemini 4 Argon
  9. 2,224 Kimi K3.1 Pro
  10. 2,120 Claude Fable 5
  11. 2,047 DeepSeek V4.1 Flash 1017
  12. 2,023 Grok 4.6
  13. 1,940 GPT-5.5 Pro
  14. 1,939 GPT-5.5
  15. 1,916 Kimi K3
  16. 1,895 GPT-5.6 Terra
  17. 1,886 space-bunny-alpha
  18. 1,872 Muse Wonder
  19. 1,858 DeepSeek V4.1 Flash
  20. 1,854 GPT-5.6 Luna
  21. 1,845 Gemini 3.7 Flash
  22. 1,809 GPT-6 Luna
  23. 1,799 Gemini 3.8 Flash
  24. 1,799 MiniMax-M3.1-Flash
  25. 1,785 Qwen 3.8 Max
  26. 1,782 GLM-5.3
  27. 1,771 Gemini 3.6 Flash
  28. 1,757 Claude Haiku 5.5
  29. 1,739 Grok 4.5
  30. 1,726 GLM-5.3-Flash
  31. 1,642 Gemini 3.1 Pro
  32. 1,640 Gemini 3.5 Flash
  33. 1,632 Claude Opus 4.8
  34. 1,627 Muse Spark 1.3
  35. 1,612 Qwen3.7-Max
  36. 1,584 DeepSeek V4 Pro 0813
  37. 1,552 GLM-5.2
  38. 1,515 Claude Sonnet 5
  39. 1,513 Muse Spark 1.1
  40. 1,472 Gemini 2.5 Deep Think
  41. 1,445 Gemini 3 Deep Think
  42. 1,443 GLM-5.1
  43. 1,426 GPT-5.2
  44. 1,401 DeepSeek V4 Flash 0731
  45. 1,396 GPT-5.4
  46. 1,394 Claude Opus 4.6
  47. 1,383 Claude Opus 4.7
  48. 1,365 MiniMax-M3
  49. 1,358 Claude Opus 4.5
  50. 1,307 GPT-5
  51. 1,289 Gemini 3 Pro
  52. 1,281 Qwen 3.6 Plus
  53. 1,275 Gemini 3 Flash
  54. 1,246 Claude Sonnet 4.5
  55. 1,241 DeepSeek-V4-Pro
  56. 1,224 GPT-5 Pro
  57. 1,167 Gemini 2.5 Pro

Bars start at 1,000.

  1. 1 GPT-6 Astra max 2,757
  2. 2 GPT-6.1 Sol max 2,647
  3. 3 Claude Opus 5.5 max 2,605
  4. 4 GPT-6 Sol max 2,426
  5. 5 Claude Opus 5 max 2,297
  6. 6 GPT-6.1 Terra max 2,295
  7. 7 GPT-5.6 Sol max 2,288
  8. 8 Gemini 4 Argon max 2,235
  9. 9 Kimi K3.1 Pro 2,224
  10. 10 Claude Fable 5 max 2,120
  11. 11 DeepSeek V4.1 Flash 1017 2,047
  12. 12 Grok 4.6 xhigh 2,023
  13. 13 GPT-5.5 Pro 1,940
  14. 14 GPT-5.5 xhigh 1,939
  15. 15 Kimi K3 max 1,916
  16. 16 GPT-5.6 Terra 1,895
  17. 17 space-bunny-alpha 1,886
  18. 18 Muse Wonder max 1,872
  19. 19 DeepSeek V4.1 Flash 1,858
  20. 20 GPT-5.6 Luna 1,854
  21. 21 Gemini 3.7 Flash 1,845
  22. 22 GPT-6 Luna max 1,809
  23. 23 Gemini 3.8 Flash 1,799
  24. 24 MiniMax-M3.1-Flash 1,799
  25. 25 Qwen 3.8 Max 1,785
  26. 26 GLM-5.3 max 1,782
  27. 27 Gemini 3.6 Flash 1,771
  28. 28 Claude Haiku 5.5 max 1,757
  29. 29 Grok 4.5 1,739
  30. 30 GLM-5.3-Flash max 1,726
  31. 31 Gemini 3.1 Pro 1,642
  32. 32 Gemini 3.5 Flash 1,640
  33. 33 Claude Opus 4.8 max 1,632
  34. 34 Muse Spark 1.3 max 1,627
  35. 35 Qwen3.7-Max 1,612
  36. 36 DeepSeek V4 Pro 0813 1,584
  37. 37 GLM-5.2 max 1,552
  38. 38 Claude Sonnet 5 xhigh 1,515
  39. 39 Muse Spark 1.1 1,513
  40. 40 Gemini 2.5 Deep Think 1,472
  41. 41 Gemini 3 Deep Think 1,445
  42. 42 GLM-5.1 1,443
  43. 43 GPT-5.2 xhigh 1,426
  44. 44 DeepSeek V4 Flash 0731 1,401
  45. 45 GPT-5.4 xhigh 1,396
  46. 46 Claude Opus 4.6 max 1,394
  47. 47 Claude Opus 4.7 max 1,383
  48. 48 MiniMax-M3 1,365
  49. 49 Claude Opus 4.5 high 1,358
  50. 50 GPT-5 high 1,307
  51. 51 Gemini 3 Pro 1,289
  52. 52 Qwen 3.6 Plus 1,281
  53. 53 Gemini 3 Flash 1,275
  54. 54 Claude Sonnet 4.5 32k thinking 1,246
  55. 55 DeepSeek-V4-Pro 1,241
  56. 56 GPT-5 Pro 1,224
  57. 57 Gemini 2.5 Pro Jun 2025 1,167
#EntryScoreSettingDateSource
1 GPT-6 Astra 2,757 max Sep 2026 Official leaderboard
2 GPT-6.1 Sol 2,647 max Sep 2026 Official leaderboard
3 Claude Opus 5.5 2,605 max Sep 2026 Official leaderboard
4 GPT-6 Sol 2,426 max Sep 2026 Official leaderboard
5 Claude Opus 5 2,297 max Jul 2026 Official leaderboard
6 GPT-6.1 Terra 2,295 max Oct 2026 Official leaderboard
7 GPT-5.6 Sol 2,288 max Jul 2026 Official leaderboard
8 Gemini 4 Argon 2,235 max Sep 2026 Official leaderboard
9 Kimi K3.1 Pro 2,224 Oct 2026 Official leaderboard
10 Claude Fable 5 2,120 max Jun 2026 Official leaderboard
11 DeepSeek V4.1 Flash 1017 2,047 Oct 2026 Official leaderboard
12 Grok 4.6 2,023 xhigh Aug 2026 Official leaderboard
13 GPT-5.5 Pro 1,940 Apr 2026 Official leaderboard
14 GPT-5.5 1,939 xhigh Apr 2026 Official leaderboard
15 Kimi K3 1,916 max Jul 2026 Official leaderboard
16 GPT-5.6 Terra 1,895 Jul 2026 Official leaderboard
17 space-bunny-alpha 1,886 Official leaderboard
18 Muse Wonder 1,872 max Oct 2026 Official leaderboard
19 DeepSeek V4.1 Flash 1,858 Sep 2026 Official leaderboard
20 GPT-5.6 Luna 1,854 Jul 2026 Official leaderboard
21 Gemini 3.7 Flash 1,845 Aug 2026 Official leaderboard
22 GPT-6 Luna 1,809 max Sep 2026 Official leaderboard
23 Gemini 3.8 Flash 1,799 Sep 2026 Official leaderboard
24 MiniMax-M3.1-Flash 1,799 Oct 2026 Official leaderboard
25 Qwen 3.8 Max 1,785 Aug 2026 Official leaderboard
26 GLM-5.3 1,782 max Aug 2026 Official leaderboard
27 Gemini 3.6 Flash 1,771 Jul 2026 Official leaderboard
28 Claude Haiku 5.5 1,757 max Oct 2026 Official leaderboard
29 Grok 4.5 1,739 Jul 2026 Official leaderboard
30 GLM-5.3-Flash 1,726 max Aug 2026 Official leaderboard
31 Gemini 3.1 Pro 1,642 Feb 2026 Official leaderboard
32 Gemini 3.5 Flash 1,640 May 2026 Official leaderboard
33 Claude Opus 4.8 1,632 max May 2026 Official leaderboard
34 Muse Spark 1.3 1,627 max Sep 2026 Official leaderboard
35 Qwen3.7-Max 1,612 May 2026 Official leaderboard
36 DeepSeek V4 Pro 0813 1,584 Aug 2026 Official leaderboard
37 GLM-5.2 1,552 max Jun 2026 Official leaderboard
38 Claude Sonnet 5 1,515 xhigh Jun 2026 Official leaderboard
39 Muse Spark 1.1 1,513 Jul 2026 Official leaderboard
40 Gemini 2.5 Deep Think 1,472 Aug 2025 Official leaderboard
41 Gemini 3 Deep Think 1,445 Feb 2026 Official leaderboard
42 GLM-5.1 1,443 Apr 2026 Official leaderboard
43 GPT-5.2 1,426 xhigh Dec 2025 Official leaderboard
44 DeepSeek V4 Flash 0731 1,401 Jul 2026 Official leaderboard
45 GPT-5.4 1,396 xhigh Mar 2026 Official leaderboard
46 Claude Opus 4.6 1,394 max Feb 2026 Official leaderboard
47 Claude Opus 4.7 1,383 max Apr 2026 Official leaderboard
48 MiniMax-M3 1,365 Jun 2026 Official leaderboard
49 Claude Opus 4.5 1,358 high Nov 2025 Official leaderboard
50 GPT-5 1,307 high Aug 2025 Official leaderboard
51 Gemini 3 Pro 1,289 Nov 2025 Official leaderboard
52 Qwen 3.6 Plus 1,281 Mar 2026 Official leaderboard
53 Gemini 3 Flash 1,275 Dec 2025 Official leaderboard
54 Claude Sonnet 4.5 1,246 32k thinking Sep 2025 Official leaderboard
55 DeepSeek-V4-Pro 1,241 Apr 2026 Official leaderboard
56 GPT-5 Pro 1,224 Oct 2025 Official leaderboard
57 Gemini 2.5 Pro Jun 2025 1,167 Jun 2025 Official leaderboard

7 models have held number one since its release in Sep 2025

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

Google DeepMind OpenAI Anthropic
Own leaderboard 7 leaders
GPT-6 Astra 2,757

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 1,500 2,000 2,500 3,000 2025 2026 2027 1,000 1,500 2,000 2,500 3,000 2025 2026 2027

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Vision

All 234 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. Kimi K3.1 Pro 84.1%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance. Command A Vision 90.9%
  • DocVQA In use Models answer 50,000 questions about 12,000 scanned document images by reading their text and layout. Scoring is ANLS, a string similarity to reference answers. Seed1.5-VL 96.9%