OpenAI research debugging - internal release score
Code
Internal research debugging evaluation with unspecified cohort and grader. Diagnostic.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Sol | 68.3 |
| GPT-5.6 Terra | 67.8 |
| GPT-5.6 Luna | 50.8 |
Benchmarks
218 sources · 1,060 models × 2,130 benchmarks · 109,661 scoresCode
Internal research debugging evaluation with unspecified cohort and grader. Diagnostic.
| Model | Score |
|---|---|
| GPT-5.6 Sol | 68.3 |
| GPT-5.6 Terra | 67.8 |
| GPT-5.6 Luna | 50.8 |