A2Z GameSpec-Bench - Source Code Big: leaderboard
Metric: Source Code Big (%; coding-agent game design fidelity; Table 2). Source: a2z-gamespec-bench.github.io. Saturation forecast: Around January 2027. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5.1 | 83.5 |
| 2 | Claude Opus 5 | 78.4 |
| 3 | GPT-6 Astra | 74.4 |
| 4 | GLM-5.3 | 58.1 |
| 5 | GPT-5.5 | 55.9 |
| 6 | DeepSeek V4 Pro | 53.8 |
| 7 | Claude Opus 4.8 | 49.6 |
| 8 | GPT-5.6 Sol | 46.3 |
Interactive version: theaggregate.ai/benchmark?slug=a2z-gamespec-bench-source-code-big · How It Works · Data refreshed daily, snapshot 2026-10-10.