Glossary
GPQA Diamond
GPQA Diamond, on the Epoch AI leaderboard (CC BY 4.0, latest version we checked), is 198 biology, physics and chemistry questions written by PhDs. A quick search almost never settles them. The score is the share of correct answers.
What it measures
GPQA Diamond, on the Epoch AI leaderboard (CC BY 4.0, latest version we checked), is 198 biology, physics and chemistry questions written by PhDs. A quick search almost never settles them. The score is the share of correct answers.
When it matters
I look at this score when the work needs technical reasoning, like reviewing a dense product memo or a calculation write-up. In this section, in the 90% to 100% range, I treat a small gap as possible measurement noise.
How to read it
Read the score as the share of questions answered correctly, on the Epoch AI leaderboard, in the latest version we checked. In this section, in the 90% to 100% range, I treat a small gap as possible measurement noise.
Who leads
- Gemini 4 Argon Google96.4%
- GPT-6 Astra OpenAI95.8%
- Claude Sonnet 5.5 Anthropic95.6%
- GPT-6.1 Sol OpenAI95.4%
- Gemini 3.8 Flash Google95.4%
- Gemini 3.7 Flash Google94.8%
- GPT-5.4 Pro OpenAI94.6%
- Gemini 3.1 Pro Google94.4%
- GPT-6 Sol OpenAI94.3%
- GPT-6.1 Terra OpenAI94.3%
Source: Epoch AI, latest version we checked.