NVDA 217.09 ▲1.38%GOOGL 344.18 ▲1.05%MSFT 533.52 ▲0.84%AMD 538.05 ▲1.76%INTC 92.03 ▲2.43%TSMC 420.22 ▲1.84%AMZN 262.45 ▲4.98%META 705.44 ▼0.55%AAPL 342.94 ▼0.15%PLTR 194.15 ▲4.36%
Markets at last close

Anthropic · Models

Claude Sonnet 5.5 leads game development ranking

·1 min read

Claude Sonnet 5.5 tops Design Arena’s game-development leaderboard with an Elo rating of 1483 in the October 2026 update. Claude Opus 5.5 follows at 1456, ahead of GPT-6 Astra at 1450. The leader’s 27-point Elo advantage over its closest rival corresponds to winning about 54 of 100 head-to-head votes.

The ranking covers 25 models and evaluates their ability to build playable games and game mechanics through human comparisons. Its focus is on functioning gameplay, including whether generated code can maintain state, physics and a game loop, rather than simply produce a convincing visual result. These scores measure performance within that specific game-development category.

GPT-6.1 Sol and Claude Fable 5.1 follow the leading group with ratings of 1386 and 1382, respectively. Kimi K3 is the highest-rated open-weight option at 1363, ranked #8 overall. It is downloadable from Hugging Face and can be self-hosted, offering an alternative for developers seeking control over deployment.

Originally reported by modelgrep.comRead the source →
Related coverage
All Anthropic news →