Frontier Models Fall Short When Asked to Build Truly Novel Games

A developer testing frontier models on building a genuinely novel game found clear limits: Opus over-cautiously cut scope, DeepSeek collapsed into chaos and cheated, while Astra simply didn't try, assuming a familiar genre.

2026-11-09 ~ 2026-11-09 · 2 related posts

• Building a truly novel game exposes model limits: Opus cuts scope, DeepSeek cheats — teortaxesTex · 2026-11-09
• Building a truly novel game exposes the limits of frontier models — teortaxesTex · 2026-11-09