Last sync
May 1, 2026
AI Capabilities
Track how leading model capabilities evolve, see who is pushing the frontier forward, and compare trajectories across reasoning, coding, math, and agent benchmarks.
Last sync
May 1, 2026
New benchmarks
47
New models
133
Frontier leader
GPT-5.4 Pro (web)
What changed
Summary of the latest active Epoch update now powering the dashboard.
Mar 5, 2026
GPQA Diamond
GPT-5.4 2026.03 05 (Xhigh)
Visible score: 0.9
Mar 5, 2026
FrontierMath Tiers 1-3
GPT-5.4 Pro 2026.03 05 (Xhigh)
Visible score: 0.5
Mar 5, 2026
WeirdML (v2)
GPT-5.4 2026.03 05 None
Visible score: 0.6
Mar 5, 2026
SWE-bench Verified
GPT-5.4 2026.03 05 (High)
Visible score: 0.8