Skip to content

Benchmark

BlenderWorld · Visual-quality dimension score

Higher scores are better.

Models and systems measured
13
Scores recorded
13
Sources
1
IQ weight
1.02

Score distribution

52.7
Mean
53.7
Median
10.1
Std dev
38.5
10th percentile
65.1
90th percentile
34.7
Min
34.772.6
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.02 for chat models, measured over 11 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1GPT-6 Astrabenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs72.6
2Claude Opus 5benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs65.1
3GPT-5.6 Solbenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs60.7
4DeepSeek V4.1 Flashbenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs59.4
5Qwen3.8 Maxbenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.55.3
6GLM 5.3 Flashbenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs54.4
7Claude Opus 4.8benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs53.7
8Qwen 3.8 27Bbenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.45.5
9Kimi K3benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs45.2
10Gemini 3.8 Flashbenchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs38.5
11MiniMax M3benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs34.7

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Grok 4.6benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs52.7
Seed2.1 Pro 0915benchmark-owner paper run · 223 tasks · Blender 4.2.10 / Blender MCP paper setup · GPT-5.6 Terra main judge · generation-agent harness and reasoning effort not stated on public leaderboard· Reported by BlenderWorld authors / Alibaba Token Hub· via BenchBoard docs47.7

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.