OpenAI GPT-5.6 Launch - Internal Research Debugging Evaluation: leaderboard
Self-improvement table of OpenAI's GPT-5.6 launch post (2026-07-09): models must find and fix real bugs in internal OpenAI research experiments that took researchers hours to days to resolve (per the GPT-5.6 system card). Score in percent, higher is better; OpenAI models only.
Metric: Score (%). Source: openai.com. Source: deploymentsafety.openai.com. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 78.05 | #6 |
| 2 | GPT-6.1 Terra | 76.39 | #15 |
| 3 | GPT-6.1 Sol | 75.52 | #7 |
| 4 | GPT-5.6 Sol | 68.3 | #22 |
| 5 | GPT-5.6 Terra | 67.8 | #40 |
| 6 | GPT-6 Sol | 64.2 | #19 |
| 7 | GPT-5.6 Luna | 50.8 | #79 |
| 8 | GPT-5.5 | 50 | #33 |
No result here: #3 Claude Opus 5.5, #9 Gemini 4 Argon, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=openai-gpt-5-6-launch-internal-research-debugging-evaluation · How It Works · Data refreshed daily, snapshot 2026-11-05.