GPT-4.5, Claude, Gemini and Grok attempt to draw the Mona Lisa

Four leading AI models — GPT-4.5, Claude, Gemini and Grok — were put to the test by attempting to recreate the Mona Lisa using coloured pencils. The experiment aimed to evaluate the visual and creative capabilities of each system. Results varied considerably across the models.

More on this topic

💻 Technology Anthropic Upgrades Claude Voice Mode With Opus and Sonnet Models 9to5Mac · 21h ago 💻 Technology AI language models show disproportionate bias towards Japanese culture PC Gamer · 1d ago 💻 Technology Cactus Post-Trains Gemma 4 to Know When It's Wrong — Cuts Cloud Queries by Up to 85% Hacker News · 1d ago 💻 Technology Google Launches Gemini 2.5 Flash, Flash-Lite and Flash Cyber AI Models Gizmodo · 2d ago 💻 Technology Google releases three new Gemini models, but 3.5 Pro still missing CNET · 2d ago 💻 Technology ChatGPT, Claude and Gemini all picked the same best smartphone Android Authority · 5d ago

More on this topic

💻 Technology Anthropic Upgrades Claude Voice Mode With Opus and Sonnet Models 9to5Mac · 21h ago 💻 Technology AI language models show disproportionate bias towards Japanese culture PC Gamer · 1d ago 💻 Technology Cactus Post-Trains Gemma 4 to Know When It's Wrong — Cuts Cloud Queries by Up to 85% Hacker News · 1d ago