promptsAI Lab

Which AI model built the best 3D solar system? Results of our test

Results of the same prompt side by side. Nothing runs on this page – click a row to open the output. Sort by clicking a column header; the best value is highlighted.

Short answer

Tested 14 Oct 2026 – 6 Nov 2026, one attempt for each of the 40 outputs, model versions are in the table. This is the result of our test, not a general measure of a model's reliability – another attempt may turn out differently. Time = from sending the prompt to the last token of the answer, thinking included (for runs in Claude Code the whole task incl. writing the file), fixes excluded. Cost = tokens × the public API price list as of the date at the model, fixes excluded; ≈ = an estimate when only the token total is known. It is not the amount actually paid – subscriptions, caching or discounts change it.

⇆ Live side by side
GPT-6.1 Sol medium
OpenAI
Model profile →
loads948/810081 / 804 min 53 s138,349$0.21
prices as of 15/10/2026
31 kB
98 lines
Claude Opus 5.5 medium
Anthropic
Model profile →
loads938/89489 / 7038 min 28 s159,363≈ $2.93
prices as of 06/10/2026
67 kB
1,210 lines
Grok 4.6 default
xAI
Model profile →
loads938/89794 / 6016 min 01 s56,405$0.33
prices as of 18/10/2026
50 kB
1,089 lines
Claude Haiku 5.5 medium
Anthropic
Model profile →
loads928/89491 / 6011 min 27 s222,948≈ $0.10
prices as of 06/10/2026
49 kB
1,268 lines
Claude Sonnet 5.5 medium
Anthropic
Model profile →
loads918/89684 / 6015 min 14 s185,039≈ $1.70
prices as of 06/10/2026
68 kB
1,189 lines
GPT-6 Astra medium
OpenAI
Model profile →
loads908/810076 / 604 min 25 s161,311$1.04
prices as of 15/10/2026
26 kB
50 lines
Kimi K3.1 Pro thinking
Moonshot AI
Model profile →
loads898/89568 / 6028 min 42 s64,268$0.64
prices as of 25/10/2026
73 kB
1,165 lines
GPT-6.1 Terra medium
OpenAI
Model profile →
loads877.5/810078 / 602 min 10 s11,954$0.058
prices as of 26/10/2026
28 kB
207 lines
GPT-5.6 Terra medium
OpenAI
Model profile →
loads847/89887 / 601 min 19 s10,517$0.12
prices as of 17/10/2026
25 kB
326 lines
DeepSeek V4.1 Flash 0910 high
DeepSeek
Model profile →
loads847/89281 / 705 min 45 s78,728$0.047
prices as of 16/10/2026
75 kB
2,089 lines
GPT-6 Sol medium
OpenAI
Model profile →
loads836.5/810079 / 802 min 27 s244,896$0.26
prices as of 15/10/2026
31 kB
83 lines
Grok 4.7 default
xAI
Model profile →
loads837/89583 / 6028 min 27 s145,007$0.86
prices as of 24/10/2026
99 kB
2,464 lines
Gemini 4 Argon medium
Google
Model profile →
loads827/87884 / 604 min 23 s29,944$0.30
prices as of 17/10/2026
72 kB
1,979 lines
DeepSeek V4.1 Flash 1017 high
DeepSeek
Model profile →
loads827/89475 / 608 min 09 s121,423$0.073
prices as of 20/10/2026
91 kB
2,655 lines
GPT-5.4 mini medium
OpenAI
Model profile →
loads806.5/89892 / 505 min 46 s92,216$0.41
prices as of 17/10/2026
67 kB
2,097 lines
GPT-5.6 Sol medium
OpenAI
Model profile →
loads766/89780 / 602 min 05 s223,637$0.33
prices as of 15/10/2026
19 kB
74 lines
Mistral Large 4 Preview none
Mistral
Model profile →
loads756.5/89680 / 3046 min 26 s120,476$0.25
prices as of 17/10/2026
72 kB
1,678 lines
Qwen3.8 Max thinking
Alibaba
Model profile →
loads746/89479 / 5024 min 32 s65,857$0.39
prices as of 22/10/2026
68 kB
1,222 lines
Hy4 preview thinking
Tencent
Model profile →
loads746/89678 / 509 min 10 s47,518$0.12
prices as of 06/11/2026
37 kB
671 lines
GPT-6 Luna medium
OpenAI
Model profile →
loads695.5/89674 / 501 min 12 s224,739$0.00811
prices as of 15/10/2026
21 kB
57 lines
Gemini 3.1 Pro medium
Google
Model profile →
loads655/88376 / 501 min 55 s19,706$0.23
prices as of 17/10/2026
39 kB
913 lines
Gemini 3.6 Flash medium
Google
Model profile →
loads645/88185 / 402 min 18 s29,794$0.11
prices as of 17/10/2026
57 kB
1,648 lines
Gemini 3.5 Flash-Lite medium
Google
Model profile →
loads603.5/89887 / 7054 s17,496$0.043
prices as of 17/10/2026
49 kB
1,199 lines
Gemma 4 31B default
Google
Model profile →
loads554/88285 / 306 min 20 s13,814$0.00
prices as of 17/10/2026
32 kB
712 lines
Mistral Medium 3.5 high
Mistral
Model profile →
loads191.5/8810 / 03 min 44 s37,053$0.28
prices as of 17/10/2026
70 kB
1,664 lines
Gemini 3.1 Flash-Lite medium
Google
Model profile →
loads120.5/8850 / 08 s3,086$0.00413
prices as of 17/10/2026
6 kB
161 lines
GPT-5.6 Luna medium
OpenAI
Model profile →
broken
fixed version →
0
after fix 7
0/8
after fix 0/8
68
after fix 68
0 / 0
after fix 0 / 0
43 s6,043$0.00687
prices as of 17/10/2026
14 kB
67 lines
GPT-5.5 medium
OpenAI
Model profile →
broken
fixed version →
0
after fix 81
0/8
after fix 6.5/8
67
after fix 96
0 / 0
after fix 91 / 60
2 min 12 s12,944$0.38
prices as of 17/10/2026
29 kB
357 lines
Gemini 3.8 Flash medium
Google
Model profile →
broken
fixed version →
0
after fix 27
0/8
after fix 1.5/8
69
after fix 65
0 / 0
after fix 0 / 60
–32,202$0.12
prices as of 16/10/2026
73 kB
2,288 lines
DeepSeek V4 Pro high
DeepSeek
Model profile →
broken
fixed version →
0
after fix 43
0/8
after fix 2.5/8
69
after fix 98
0 / 0
after fix 94 / 0
4 min 22 s41,688$0.082
prices as of 16/10/2026
88 kB
2,060 lines
Mistral Small 4 high
Mistral
Model profile →
broken
fixed version →
0
after fix 18
0/8
after fix 1.5/8
64
after fix 64
0 / 0
after fix 0 / 0
2 min 15 s27,061$0.016
prices as of 17/10/2026
42 kB
1,156 lines
Codestral 25.08 default
Mistral
Model profile →
broken
fixed version →
0
after fix 7
0/8
after fix 0/8
67
after fix 67
0 / 0
after fix 0 / 0
2 min 58 s21,146$0.019
prices as of 17/10/2026
90 kB
2,133 lines
GLM 5.3 high
Z.ai
Model profile →
broken
fixed version →
0
after fix 26
0/8
after fix 2.5/8
70
after fix 71
0 / 0
after fix 0 / 0
6 min 34 s101,405$0.45
prices as of 17/10/2026
57 kB
1,215 lines
Kimi K3 thinking
Moonshot AI
Model profile →
broken
fixed version →
0
after fix 92
0/8
after fix 8/8
66
after fix 95
0 / 0
after fix 91 / 60
37 min 14 s64,889$0.97
prices as of 22/10/2026
55 kB
1,064 lines
Muse Wonder thinking
Meta
Model profile →
broken
fixed version →
0
after fix 83
0/8
after fix 7/8
–
after fix 96
– / 0
after fix 79 / 60
2 min 03 s20,212$0.40
prices as of 25/10/2026
41 kB
367 lines
MiniMax M3.1 Flash thinking
MiniMax
Model profile →
broken
fixed version →
0
after fix 81
0/8
after fix 7/8
–
after fix 95
– / 0
after fix 66 / 60
6 min 39 s74,914$0.045
prices as of 25/10/2026
92 kB
2,153 lines
MiMo-V2.6-Pro thinking
Xiaomi
Model profile →
broken
fixed version →
0
after fix 76
0/8
after fix 6.5/8
94
after fix 93
76 / 0
after fix 79 / 40
23 min 30 s82,126$0.071
prices as of 25/10/2026
80 kB
1,941 lines
Qwen3.6 27B thinking
Alibaba
Model profile →
broken
fixed version →
0
after fix 7
0/8
after fix 0/8
–
after fix 68
– / 0
after fix 0 / 0
4 min 47 s20,815$0.041
prices as of 25/10/2026
49 kB
1,328 lines
Mistral Large 4 1029 medium
Mistral
Model profile →
broken
fixed version →
0
after fix 7
0/8
after fix 0/8
68
after fix 68
0 / 0
after fix 0 / 0
2 min 15 s18,521$0.038
prices as of 31/10/2026
34 kB
320 lines
GLM-5.3-Flash high
Z.ai
Model profile →
broken
fixed version →
0
after fix 75
0/8
after fix 6/8
70
after fix 100
0 / 0
after fix 75 / 60
8 min 12 s80,922$0.040
prices as of 31/10/2026
57 kB
1,094 lines

Cost = tokens priced with the API price list valid on the date shown next to the cost (prices change). ≈ = estimate (only the total token count is known), no ≈ = exact from input / output tokens. “Loads” = the page opens without errors; technical health also checks behaviour during use – judge the quality in the viewer.

Evaluation

40 outputs from 40 models for the same zero-shot prompt – each one the model's first answer, published unedited.

  • 4/8average score
  • 26/40working outputs
  • 86/100avg. technical health
  • 55/100avg. UX

Broken (do not run): 14, 14 of them with a copy fixed by the same model. The average checklist score is 4/8 (a broken output scores 0), 6.2/8 among working ones. Best result in our test: Claude Opus 5.5, Claude Haiku 5.5, Claude Sonnet 5.5 and 4 more – 8/8. Weakest working: Gemini 3.1 Flash-Lite – 0.5/8.

How we evaluate – methodology →

How the 8 requirements were met

From 26 working outputs: ✓ met, ½ partly, ✗ missing. Rate = (✓ + ½ × 0.5) / number of outputs.

Requirement✓½✗Rate
3D scene
3D scene: Sun, 8 planets, Moon, asteroid belt, stars
1781 81 %
Recognizable planets (Jupiter bands & spot, Saturn rings, Earth oceans & clouds), sunlight 13121 73 %
Real data
Real data: periods, eccentricity & inclination, positions roughly right for the date
1592 75 %
Time
Time: play/pause, reverse, real time to years/s, “today”, date shown
1781 81 %
Camera
Camera: orbit, zoom and pan with mouse and touch
1592 75 %
Click / list
Click / list: smooth fly-to, follow, info panel
1772 79 %
Orbit & label toggles, readable vs true scale switch 13112 71 %
Single file, no external assets, full window, resize & high-DPI, runs without errors 2042 85 %

Strengths

The requirements the models met best – quotes from the reviewer's verdicts:

Single file, no external assets, full window, resize & high-DPI, runs without errors · 85 % met

  • Single file, no assets, resize and high-DPI handled, renders fine. – Claude Opus 5.5
  • Single self-contained file, resize and high-DPI handled. – Claude Haiku 5.5

3D scene · 81 % met

  • Sun, eight planets, Moon, asteroid belt and stars all present. – Claude Opus 5.5
  • Sun, eight planets, Moon, asteroid belt and starfield present. – Claude Haiku 5.5

Typical mistakes

The requirements the models most often missed or met only partly:

Orbit & label toggles, readable vs true scale switch · 71 % met

  • Labels missing, orbits invisible, scale switch barely works – Mistral Medium 3.5
  • Toggles do nothing; no scale switch. – Gemini 3.1 Flash-Lite

Recognizable planets (Jupiter bands & spot, Saturn rings, Earth oceans & clouds), sunlight · 73 % met

  • No visible planets or planet details. – Gemini 3.1 Flash-Lite
  • Good textures and lighting, but spheres render with a hole grid. – DeepSeek V4.1 Flash 1017

Camera · 75 % met

  • Camera is coded, but the frozen page does not respond. – Mistral Medium 3.5
  • Camera cannot be controlled. – Gemini 3.1 Flash-Lite

Agreement of two independent reviewers

The checklist of 41 outputs was filled in independently by two AI reviewers, Claude Opus 5.5 and Codex (GPT-6 Astra, medium). They agreed on 274 of 328 items (84 %). Average score by Claude 5.1/8, by Codex 5/8. Codex gave a lower score for 19, a higher one for 8 and the same for 14 outputs. They disagreed most often on: Orbit & label toggles, readable vs true scale switch (11×), Click / list (10×), Single file, no external assets, full window, resize & high-DPI, runs without errors (7×).

Where the reviewers disagree, the stricter (lower) verdict is proposed; the published score is approved by a human.

Objective tests

Measured by a tool, not by a person or an AI – so they are shown even before the scores are approved.

  • Tech: average 86/100 from 37 measurements; highest GPT-6 Astra (100), lowest Mistral Small 4 (64). Measures Lighthouse, smoothness, errors during automated use, the mobile layout and whether the controls work.
  • UX / UI: average 55/100 from 37 measurements; highest Grok 4.6 (94), lowest GLM-5.3-Flash (0). Measures control size, layout at 3 widths, keyboard, feedback and readability.