I Made 3 AI Models Compete to Make a Videogame

I made three major AI models compete to build a complete three-level platformer from the exact same prompt, using only HTML, CSS, and vanilla JavaScript inside a single file. The three competitors, ChatGPT 5.5 Medium, Gemini Flash Thinking, and Claude Sonnet 5 Medium, were scored across functionality, gameplay, design, presentation, and reliability. Then I gave the same prompt to the new GPT-5.6 Sol Ultra as a completely unscored bonus round, and the result was honestly unbelievable! Play every game yourself, decide whether you agree with my rankings, and tell me what score Sol Ultra deserves out of 100! My Links 🔗 ➡️ Play all four games and view the code: https://github.com/lerabyte/ai-platfo... Open the repository, go directly to the README, and click “Click here to open the game hub.” From there, you can launch and play all four platformers directly in your browser. Please star the repository if you enjoy the project! ⭐ ➡️ TikTok:   / lerabyte   ➡️ Instagram:   / lera_byte   ➡️ LinkedIn:   / lera-andronova-a5422029b   My Scores ChatGPT 5.5 Medium Game: Lumenfall: Echoes of the Skyforge Final score: 96/100 Gemini Flash Thinking Game: Neon Cyber-Run Final score: 88/100 Claude Sonnet 5 Medium Game: LUMEN: Crystal Caverns Final score: 93/100 GPT-5.6 Sol Ultra Game: Lumenling: The Shattered Skygarden Final score: You decide! If you encounter anything interesting while playing, discover a bug, find a hidden detail, or have an idea for improving one of the games, please report it in the comments. I also want to know what challenge I should give these models next! Timestamps ⏱️ 03:09 ChatGPT 5.5 Medium 08:45 Gemini Flash Thinking 14:17 Claude Sonnet 5 Medium 18:53 GPT-5.6 Sol Ultra Bonus Round #AI #MachineLearning #ChatGPT #Gemini #Claude #Coding #GameDevelopment