VoxelBench (text)

Models build 3D voxel scenes from text prompts; humans vote head-to-head on the renders. Glicko-2 rating.

54 models · 54 results · data as of 2026-10-29 · voxelbench.ai ↗

All results

54 rows

#	Model	Effort	VoxelBench (text)
1	GPT-6.1 Sol	max	2698 ±210
2	GPT-6 Astra	max	2677 ±245
3	Opus 5.5	max	2625 ±112
4	GPT-6 Sol	max	2417 ±108
5	GPT-6.1 Terra	max	2341 ±137
6	GPT-5.6 Sol	max	2282 ±96
7	Opus 5	max	2259 ±102
8	Kimi K3.1	max	2236 ±65
9	Gemini 4 Argon	max	2234 ±94
10	Fable 5	max	2105 ±102
11	Grok 4.6	xhigh	2070 ±98
12	DeepSeek V4.1 Flash 1017	max	2057 ±90
13	GPT-5.5 Pro	xhigh	1933 ±100
14	GPT-5.5	xhigh	1924 ±96
15	Kimi K3	max	1917 ±94
16	GPT-5.6 Terra	max	1902 ±94
17	DeepSeek V4.1 Flash	max	1866 ±100
18	GPT-5.6 Luna	max	1828 ±94
19	Gemini 3.7 Flash	max	1827 ±96
20	GPT-6 Luna	max	1819 ±102
21	Gemini 3.8 Flash	max	1807 ±100
22	MiniMax M3.1	max	1805 ±86
23	Qwen 3.8 Max	max	1801 ±94
24	GLM 5.3	max	1767 ±92
25	Gemini 3.6 Flash	xhigh	1759 ±92
26	Grok 4.5	max	1758 ±94
27	Haiku 5.5	max	1744 ±96
28	GLM 5.3 Flash	max	1710 ±92
29	Muse Spark 1.3	max	1649 ±108
30	Gemini 3.1 Pro	high	1648 ±98