Capabilities · Trend
Efficient small models
Distilled, smaller models match last year's frontier — capability per watt is the new race.
Forecast horizon
On-device and private deployment spread; sovereignty and offline use become realistic.
Connections
Connections · 4
How this node ties into the rest of the map, and the evidence behind each link.
to · drives 7/10
Collapsing inference cost Smaller models slash cost per capability.
+3 growth to · enables 6/10
AI in public services Efficient models make on-premise public deployment feasible.
+2 growth from · supports 6/10
Open-weight models Open releases seed efficient downstream models.
+2 growth from · enables 5/10
Synthetic data Distillation via synthetic data shrinks capable models.
+2 growthSignal sources
Signal sources
Dated facts from primary sources in this direction.
Task horizon doubling Mar 2025
The length of software tasks AI agents can do autonomously at 50% reliability has doubled about every 7 months — and since 2024 closer to every ~3 months.
METR → Benchmarks saturating Apr 2025
In one year scores rose by 18.8, 48.9 and 67.3 points on MMMU, GPQA and SWE-bench; real-world software solve rate jumped from 4.4% to 71.7%.
Stanford HAI — AI Index 2025 → Autonomous coding 2025–2026
On SWE-bench Verified (500 real GitHub issues), autonomous coding agents reached ~80–86% by late 2025, up from under 50% in early 2025.
Epoch AI →