OpenAI GPT-5.5 System Card - Hard Negative Protein Binding Prediction: leaderboard

Biosecurity capability slice from OpenAI's GPT-5.5 system card — hard-negative protein-binding prediction; OpenAI-run frontier comparison.

Metric: pass@4 (%). Source: deploymentsafety.openai.com. 3 models tracked.

Top models

#ModelScoreOverall rank
1GPT-5.4 (Thinking)3.46#79 (GPT-5.4)
2GPT-5.51.48#31
3GPT-5.5 Pro0#17

Interactive version: theaggregate.ai/benchmark?slug=openai-gpt-5-5-system-card-hard-negative-protein-binding-prediction · How It Works · Data refreshed daily, snapshot 2026-10-20.