4DCodeBench: leaderboard
Metric: Reconstruction score (0–1), five-family mean over 200 scenes; provider-specific coding harnesses. Source: 4dcodebench.com. Saturation forecast: Around April 2027. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 Astra (Max) | 0.79 |
| 2 | Claude Opus 5.5 (High) | 0.78 |
| 3 | GPT-6 Astra (High) | 0.77 |
| 4 | Claude Fable 5.1 (High) | 0.74 |
| 5 | GPT-6 Astra (Low) | 0.73 |
| 6 | Claude Opus 5 (High) | 0.72 |
| 7 | GPT-5.6 Luna (Max) | 0.67 |
| 8 | GPT-5.6 Sol (High) | 0.65 |
| 9 | Gemini 3.8 Flash (High) | 0.64 |
| 10 | DeepSeek V4.1 Flash (High) | 0.6 |
| 11 | GPT-5.6 Terra (High) | 0.55 |
| 12 | GLM-5.3 Flash (Max) | 0.47 |
| 13 | Muse Glimmer 30B (High) | 0.44 |
| 14 | MiMo-V2.5 | 0.43 |
| 15 | MiniMax-M3 | 0.34 |
Interactive version: theaggregate.ai/benchmark?slug=4dcodebench · How It Works · Data refreshed daily, snapshot 2026-10-08.