Visual / Visual Abstraction

ASCII Mona Lisa

A tiny prompt asks the model to draw the Mona Lisa using only text characters. The portrait should be recognizable before the page title or frame gives it away.

Prompt

Create single-HTML ASCII art of the Mona Lisa

Max tokens
100K
temperature
0
top_p
1
seed
42
presence_penalty
0
frequency_penalty
0
Reasoning effort
High
Execution
Single-shot via API

Fortytwo Prime

Fortytwo

PASS2 / 2

Fortytwo reads as the Mona Lisa at a glance, with coherent face, hair, shoulders, and pose, strong tonal character density, and polished page presentation.

vs
Gemma 4Google
Claude Fable 5AnthropicMIXEDOpenRouter Fusion (Quality)OpenRouterFAILOpenRouter Fusion (Budget)OpenRouterMIXEDKimi K2.7 CodeMoonshotFAILMiniMax M3MiniMaxFAILClaude Opus 4.8AnthropicFAILDeepSeek V4 ProDeepSeekFAILChatGPT 5.5OpenAIFAILGrok 4.3xAIFAILClaude Opus 4.7AnthropicFAILQwen 3.6 PlusAlibabaFAILGrok 4.20xAIFAILMiniMax M2.7MiniMaxFAILChatGPT 5.4OpenAIFAILClaude Opus 4.6AnthropicFAILGemini 3.5 FlashGoogleFAILGemma 4GoogleFAILGemini 3.1 ProGoogleFAILNemotron 3 UltraNVIDIAFAILDeepSeek V4 FlashDeepSeekFAILGLM 5.1ZhipuFAILKimi K2.5MoonshotFAILDeepSeek V3.2DeepSeekFAIL
FAIL0 / 2

Gemma 4 uses character density, but the cropped generic silhouette and underdefined face do not read as the Mona Lisa, and the presentation is awkwardly cropped enough to hurt legibility.

Model verdicts

ModelVerdictRecognizable PortraitPresentation Quality
1Fortytwo PrimePASSpasspass2Claude Fable 5MIXEDmixedpass3OpenRouter Fusion (Budget)MIXEDmixedmixed4OpenRouter Fusion (Quality)FAILfailpass5Kimi K2.7 CodeFAILfailpass6MiniMax M3FAILfailpass7Claude Opus 4.8FAILfailpass8DeepSeek V4 ProFAILfailpass9ChatGPT 5.5FAILfailpass10Grok 4.3FAILfailpass11Claude Opus 4.7FAILfailpass12Qwen 3.6 PlusFAILfailpass13Grok 4.20FAILfailpass14MiniMax M2.7FAILfailpass15ChatGPT 5.4FAILfailpass16Claude Opus 4.6FAILfailpass17Gemini 3.5 FlashFAILfailmixed18Gemma 4FAILfailmixed19Gemini 3.1 ProFAILfailmixed20Nemotron 3 UltraFAILfailfail21DeepSeek V4 FlashFAILfailfail22GLM 5.1FAILfailfail23Kimi K2.5FAILfailfail24DeepSeek V3.2FAILfailfail

Have a complex task to evaluate?

Request a custom evaluation for your use case.

Request a demo →