Updated
Which AI model writes the most code-quality issues?
As of Nov 14, 2026, GPT-6 Astra had the highest rate of code-quality issues (1.26, 26% above the all-model rate), statistically tied with Claude Fable 5.1. GPT-6.1 Sol had the lowest (0.69, 31% below average), statistically tied with Claude Opus 5.5 and Claude Fable 5.1.
| Rank | Model | Rate |
|---|---|---|
| 1= (statistically tied) | GPT-6.1 Sol | 0.69 |
| 2= (statistically tied) | Claude Opus 5.5 | 0.74 |
| 3= (statistically tied) | Claude Fable 5.1 | 0.92 |
| 4= (statistically tied) | GPT-6 Astra | 1.26 |
1.00 = all-model rate · lower is better · = statistically tied