Reasoning and knowledge
GPQA Diamond
1 model with a published score, best first. Higher is better.
96.0%GPT-6 AstraRank 1, published Oct 20, 2026
Leaderboard
Each model's latest published score. Every score links to its source.
| # | Model | GPQA | Date | Note | Source |
|---|---|---|---|---|---|
| 1 | O GPT-6 AstraRecordOpenAI | 96.0% | Oct 20, 2026 | provider-reported | benchlm.ai |
Record history
The best score ever published on GPQA Diamond, and who held it. Each step is a new record.
| Date | Record holder | GPQA | Gain | Note |
|---|---|---|---|---|
| Oct 20, 2026 | GPT-6 AstraCurrent | 96.0% | First | provider-reported |
Method
Close to saturated. Higher is better. Each score links to its own source.
We collect the scores the board and the labs publish and run none ourselves. Ranks use each model's latest reading. 1 reading stored for this board, the newest from Oct 20, 2026.