Release Analytics
Cumulative Releases Over Time
The compounding pace of frontier model releases.
Release Frequency (Per Month)
New models shipped each calendar month.
Releases by Year
Year-over-year totals. 2026 is partial (through May 18).
GPQA Diamond — All Models by Company
Every published GPQA Diamond score, plotted by release date and colored by lab. The frontier emerges at the top edge.
GPQA Diamond — Best Score Frontier
The highest GPQA Diamond score achieved by any model, month over month. The frontier only ever moves up.
Open-Source vs Closed — GPQA Frontier
Best GPQA Diamond score over time, split by license. Tracks how fast open-weight models are catching up to proprietary ones.
SWE-Bench Verified — Best Score Frontier
How well an AI can fix real bugs in real software. It's given a bug report from an actual project and has to write the code that fixes it. Higher score = better programmer.
Top 5 Models by GPQA Diamond
Hard PhD-level science questions in physics, biology, and chemistry — written so you can't just google the answer. For reference: PhDs in the field score ~65%.
Release Cadence
Average number of days between releases per company. Shorter bars = ships more often.