I put Opus 5.5, GPT-6 Astra and Fable 5.1 through the same game-building challenges, from underwater Atlantis to brick-built racing, then compared what each one made.
Which AI would you trust to build your next game?
I put Opus 5.5, GPT-6 Astra and Fable 5.1 through a series of game-building challenges, from underwater Atlantis exploration to brick-built racing and fantasy worlds.
We’re looking at the actual results: graphics, gameplay, world detail and how the games feel to play. Beautiful screenshots are one thing. Building something you actually want to explore is another.
Which model would you choose, and what should I make them build next?
How far can you push Opus 5.5 for AI game development? I challenged it with an Elden Ring-inspired world, a Far Cry-style jungle shooter, and a sprawling voxel fantasy town. In this video, I explor…
Watch the update
What happens when you give Sonnet 5.5 and GPT-6.1 SOL several hours to build and improve a game? I ran both models through repeated prompts and rounds of revisions to see how far they could go. Thi…
Watch the update