Charts

All charts

Daily-tracked signals from across the AI capability race — model trajectories, enterprise routing shifts, benchmark events, training compute. New charts added as the data moves.

April 2026 Capabilities

METR Projections — Two Labs, Two Trajectories

METR Projections — Two Labs, Two Trajectories
What it shows: 50% task-completion horizon (METR), log scale, fit per lab. Anthropic broke trend at Sonnet 4.5 — doubling every 1.7 months since. OpenAI holds steady at 3.8. Mythos forecast: ~28h. GPT-5.5: ~9h.