Models
Every generator in the arena, grouped by the method it uses — a model is only ranked against the others in its own section. BT scores come from separate match pools per method, so they aren't comparable across sections. Click through for per-task stats.
Image→3D reconstruction
12 ranked of 12 · view the Image→3D reconstruction board →One or more photos reconstructed into a 3D mesh.
#1
Meshy 6
model
Image→3D reconstruction
1145.7 BT in method
142 votes
11 tasks
#1
TRELLIS 2
model
Image→3D reconstruction
1067.1 BT in method
57 votes
6 tasks
#2
Hunyuan3D v3
model
Image→3D reconstruction
1044.0 BT in method
115 votes
12 tasks
#2
Rodin/Hyper3D
model
Image→3D reconstruction
1014.7 BT in method
91 votes
11 tasks
#2
Pixal3D
model
Image→3D reconstruction
1011.1 BT in method
125 votes
11 tasks
#2
Hunyuan3D 3.1
model
Image→3D reconstruction
986.8 BT in method
112 votes
12 tasks
#2
Hunyuan3D v2
model
Image→3D reconstruction
982.9 BT in method
95 votes
11 tasks
#4
SAM 3D
model
Image→3D reconstruction
941.1 BT in method
158 votes
15 tasks
#4
TRELLIS via Replicate
model
Image→3D reconstruction
907.6 BT in method
75 votes
11 tasks
#4
TRELLIS via fal
model
Image→3D reconstruction
878.4 BT in method
62 votes
8 tasks
#4
TripoSR
model
Image→3D reconstruction
834.0 BT in method
42 votes
7 tasks
#2
InstantMesh
model
Image→3D reconstruction
729.1 BT in method
provisional
1 votes
1 tasks
LLM procedural (code-gen)
19 ranked of 19 · view the LLM procedural (code-gen) board →An LLM writes code (e.g. Blender) that builds the 3D model.
anthropic/claude-opus-4.8
model
commissioned via OpenRouter
1012.9 BT in method
provisional
12 votes
12 tasks
#1
anthropic/claude-sonnet-5
model
commissioned via OpenRouter
1012.9 BT in method
provisional
15 votes
13 tasks
deepseek/deepseek-v3.2
model
commissioned via OpenRouter
1012.9 BT in method
provisional
16 votes
14 tasks
deepseek/deepseek-v4-pro
model
commissioned via OpenRouter
1012.9 BT in method
provisional
9 votes
10 tasks
google/gemini-3.1-pro-preview
model
commissioned via OpenRouter
1012.9 BT in method
provisional
8 votes
11 tasks
meta-llama/llama-4-maverick
model
commissioned via OpenRouter
1012.9 BT in method
provisional
9 votes
6 tasks
minimax/minimax-m3
model
commissioned via OpenRouter
1012.9 BT in method
provisional
7 votes
8 tasks
mistralai/mistral-medium-3-5
model
commissioned via OpenRouter
1012.9 BT in method
provisional
10 votes
12 tasks
moonshotai/kimi-k2.7-code
model
commissioned via OpenRouter
1012.9 BT in method
provisional
9 votes
10 tasks
openai/gpt-5.1
model
commissioned via OpenRouter
1012.9 BT in method
provisional
3 votes
6 tasks
openai/gpt-5.6-sol
model
commissioned via OpenRouter
1012.9 BT in method
provisional
14 votes
14 tasks
openai/gpt-5.6-sol-pro
model
commissioned via OpenRouter
1012.9 BT in method
provisional
1 votes
1 tasks
qwen/qwen3.6-plus
model
commissioned via OpenRouter
1012.9 BT in method
provisional
4 votes
4 tasks
qwen/qwen3.7-plus
model
commissioned via OpenRouter
1012.9 BT in method
provisional
10 votes
13 tasks
x-ai/grok-4.20
model
commissioned via OpenRouter
1012.9 BT in method
provisional
14 votes
14 tasks
x-ai/grok-4.3
model
commissioned via OpenRouter
1012.9 BT in method
provisional
4 votes
4 tasks
#1
x-ai/grok-4.5
model
commissioned via OpenRouter
1012.9 BT in method
provisional
9 votes
13 tasks
#1
z-ai/glm-4.6v
model
commissioned via OpenRouter
1012.9 BT in method
provisional
8 votes
8 tasks
z-ai/glm-5.2
model
commissioned via OpenRouter
1012.9 BT in method
provisional
12 votes
12 tasks
Text→3D (native)
7 ranked of 7 · view the Text→3D (native) board →A text prompt turned directly into 3D.
#1
Meshy v6 text
model
Text→3D (native)
1086.4 BT in method
219 votes
14 tasks
#1
Tripo H3.1 text
model
Text→3D (native)
1064.3 BT in method
242 votes
16 tasks
#1
Tripo P1 text
model
Text→3D (native)
1038.3 BT in method
229 votes
15 tasks
#3
Rodin text via fal
model
Text→3D (native)
976.7 BT in method
218 votes
14 tasks
#4
Rodin text via Replicate
model
Text→3D (native)
958.7 BT in method
145 votes
11 tasks
#4
Hunyuan3D 3.1 text
model
Text→3D (native)
951.2 BT in method
195 votes
13 tasks
#4
Hunyuan3D v3 text
model
Text→3D (native)
930.4 BT in method
241 votes
16 tasks
Agentic 3D
15 ranked of 15 · view the Agentic 3D board →An agent renders, critiques, and revises its own 3D model in a loop.
anthropic/claude-opus-4.8 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
15 votes
11 tasks
anthropic/claude-sonnet-5 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
11 votes
9 tasks
google/gemini-3.1-pro-preview (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
14 votes
12 tasks
meta-llama/llama-4-maverick (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
6 votes
6 tasks
minimax/minimax-m3 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
6 votes
6 tasks
mistralai/mistral-medium-3-5 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
5 votes
4 tasks
moonshotai/kimi-k2.7-code (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
4 votes
4 tasks
openai/gpt-5.1 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
7 votes
6 tasks
openai/gpt-5.6-sol (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
13 votes
12 tasks
openai/gpt-5.6-sol-pro (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
5 votes
4 tasks
qwen/qwen3.6-plus (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
1 votes
1 tasks
qwen/qwen3.7-plus (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
8 votes
8 tasks
x-ai/grok-4.20 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
8 votes
7 tasks
x-ai/grok-4.5 (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
9 votes
9 tasks
z-ai/glm-4.6v (agentic)
model
agentic (iterative render-critique-revise) via OpenRouter
1012.9 BT in method
provisional
9 votes
8 tasks