OpenAI GPT-5.6 System Card - HealthBench Hard: leaderboard
Hard subset of HealthBench, the conversations frontier models still fail, on the length-adjusted scale; OpenAI's own runs as published in the GPT-5.6 system card.
Metric: Length-Adjusted Score (0-100). Source: deploymentsafety.openai.com. Saturation forecast: Around 2030. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 34.7 | #98 |
| 2 | GPT-5.2 | 34.3 | #110 |
| 3 | GPT-5.6 Sol | 33.1 | #19 |
| 4 | GPT-5.6 Terra | 32.7 | #41 |
| 5 | GPT-5.6 Luna | 32 | #68 |
| 6 | GPT-5.5 | 31.5 | #31 |
| 7 | GPT-5.4 | 29.1 | #79 |
| 8 | GPT-5.1 | 25.4 | #135 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openai-gpt-5-6-system-card-healthbench-hard · How It Works · Data refreshed daily, snapshot 2026-10-20.