Output Loop Bench · Marketing

Plumber flyer (A5)

Why we chose this prompt: A print piece a tradesperson would really hand out. It shows whether a model can build clear hierarchy, an offer and trust signals that read at arm's length.

ModelMakerTimeTokens
claude-opus-5-5 ★ Output Loop's pickAnthropic181.1s106,117
muse-spark-1.3-mediumMeta16.1s24,842
nemotron-3-nano-4bNVIDIA16.5s1,984
composer-2.5Cursor19.1s22,945
gemini-3.8-flashGoogle32.8s21,627
claude-haiku-5-5Anthropic41.4s99,598
gemma-4-26b-a4b-qatGoogle45.3s5,718
grok-4.7-build-fastxAI49.6s23,967
gemini-4-argon-mediumGoogle53.0s22,201
kimi-k3-maxMoonshot (Kimi)54.3s29,047
glm-5.3Z.ai59.3s19,269
qwen3.8-flashAlibaba (Qwen)59.7s21,568
deepseek-v4-proDeepSeek65.1s22,927
muse-glimmer-30bMeta65.3s18,631
MiniMax-M2.5MiniMax66.0s21,820
gpt-6-astraOpenAI75.3s17,625
gpt-6.1-solOpenAI91.8s18,016
kimi-k2.7-codeMoonshot (Kimi)92.0s39,194
qwen3.7-plusAlibaba (Qwen)96.1s22,073
claude-fable-5-1Anthropic102.8s97,797
claude-sonnet-5-5Anthropic117.1s105,726
muse-glimmerMeta122.4s4,154
bonsai-27bPrism ML129.3s7,305
gemini-3.1-proGoogle130.2s22,642
deepseek-v4.1-flashDeepSeek170.9s36,559
kimi-k3Moonshot (Kimi)282.5s17,849
gpt-6-lunaOpenAI293.4s30,815
grok-4.7xAI340.7s43,942
qwen3.8-maxAlibaba (Qwen)527.6s40,012
nemotron-3.5-lightning-30b-a3bNVIDIA552.5s222,755
qwen3.8-27bAlibaba (Qwen)980.9s30,786

Open this task in the interactive Bench