Arena AI: Compare AI image models on LM Arena

Demo summary

The user demonstrates generating images on the LM Arena platform to compare outputs from different models like Kwen Image and Flux Pro side-by-side.

Navigate to the LMSYS Chatbot Arena and click on New Chat
Select the Generate Images option
Paste your image prompt into the text field
Click on Generate
Log in with a Google account to proceed with the generation
Compare the side-by-side results from the two different models
Vote for the better image by selecting A is better, B is better, Both are good, or Both are bad
Download the preferred image

Use a prompt that includes text to test how different models handle typography
Compare the specific details of the prompt, such as motion blur, to determine which model followed instructions more accurately

“I specifically love this AI tool as it gives us, you know, the best to choose from while using two completely different AI image models.”

3:521:18Generate and edit images with Meta AIThe creator uses Meta AI to generate a high-quality image from a Claude-assisted prompt and demonstrates the 'Add a black hat' editing feature to modify the image without credits.AI at Meta· AI Inpainting
6:071:26Animate images with Meta AIThe video shows how to use Meta AI's 'Animate' and 'Custom Animate' features to turn static images into video clips with specific motion prompts.AI at Meta· Image to Video
8:231:21Compare AI image models on LM ArenaCurrentThe user demonstrates generating images on the LM Arena platform to compare outputs from different models like Kwen Image and Flux Pro side-by-side.Arena AI· AI Image Generator
9:551:16Generate videos with Seedance and Happy HorseThe creator uses the LM Arena video generation interface to produce and compare 8-second video clips using the Seedance 1.5 and Happy Horse models.Arena AI· Text to Video
12:021:12Text-to-image and image-to-video on Vidu AIThe video walks through using Vidu AI (referred to as Vhere AI in captions) to generate a dog image and then animate a photo of a woman into a realistic video.Vidu· Image to Video
14:381:32Generate realistic video with CapCut DreaminaThe creator demonstrates using the Seedance 2 model within CapCut Dreamina to generate a realistic video of a woman walking in New York with zero credit cost.Dreamina· Text to Video
Watch “3 FREE AI Video Generators That Are Actually UNRESTRICTED” →

1:341:29Compare AI image models on LM ArenaThe user demonstrates how to use the 'Battle Mode' in LM Arena to generate images from a text prompt and compare outputs from anonymous models, eventually revealing Grok as the winner.Manny ai
8:231:21Compare AI image models on LM ArenaCurrentThe user demonstrates generating images on the LM Arena platform to compare outputs from different models like Kwen Image and Flux Pro side-by-side.Brain Project
1:273:43Generate and compare images in Arena.ai Battle ModeThe user demonstrates entering a text prompt into Arena.ai's anonymous battle mode to generate two images from different unknown models and then votes on the preferred result.Ai Atomation
5:104:08Side-by-side image generation with specific modelsThe presenter shows how to manually select specific AI models (GPT and Gemini) within Arena.ai to generate and compare images for the same prompt simultaneously.Ai Atomation
2:400:30Compare image generation modelsA prompt for a monster climbing a building is used to compare two image models side-by-side, identifying Imagen 4 from Google as the winner.Eric Michaud