JamBench - Theme-Only Generation (Claude Code): leaderboard
Metric: L3a runtime pass rate (%): share of generated Godot 4 projects that pass file-integrity and headless compilation checks and then run 30 seconds in headless mode without crashing; Task 1a: only the jam theme keyword is given; 50 real Game Jam themes, three runs each; the model writes a complete 2D Godot 4 project from scratch; Claude Code as the agent harness, iterating with the Godot toolchain; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V4 Pro | 84 |
| 2 | Claude Opus 4.6 | 82.7 |
| 3 | Kimi K2.5 | 81.3 |
| 4 | Qwen 3.5 397B A17B | 80.7 |
| 5 | GLM-5 | 78.7 |
Interactive version: theaggregate.ai/benchmark?slug=jambench-theme-only-generation-claude-code · How It Works · Data refreshed daily, snapshot 2026-09-29.