The absurd pelican testSVG makes the pelican test workAmir Bohlooli / MUODeveloper and AI researcher Simon Willison introduced the test on October 25, 2024. Even big frontier models struggle with the pelicanGemini won, but not cleanlyScreenshot by Amir Bohlooli for MUOI tested the prompt on Claude, Gemini, and ChatGPT. Image by Amir Bohlooli for MUO Image by Amir Bohlooli for MUOYou can see the results in the galleries I've attached. Local models aren’t that far behindIn fact, I like them betterScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOClose Screenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOScreenshot by Amir Bohlooli for MUOGemini certainly produced the most impressive result among the frontier models, but it also added a lot that I never asked for. The local models don’t have a per-token bill ticking upward while I use them, although the hardware and electricity certainly aren’t free.