Bills Went Up After OpenAI's 80% Price Cut: Jevons Paradox Hits Inference
AccBalanced · x · 2026-08-18
Via @meggmcnulty: when OpenAI cut Luna pricing by 80% last month, most companies didn't save money — teams built more ambitious agent workflows because the inference constraint relaxed, and total bills rose materially.
The reason: the harness (orchestration deciding retries, context pulls, tool-call budgets) burned more tokens per task because each token felt cheap. And nobody measured cost per completed task, so the metric that mattered stayed flat or worsened while the tracked one fell off a cliff. This is Jevons paradox applied to inference — in 1865 Jevons observed that coal efficiency gains increased total coal consumption. Expect the same in 2026-2027 enterprise AI budgets.
More from Infra
- Model Prices Collapse, but Gateway Fees Remain High — AccBalanced · 2026-08-18
- Performance Analysis of Mixing RTX 3080 and A2000 for Stable Diffusion — djekler · 2026-08-18
- Snapdragon X2 Elite Extreme Doubles AI Performance Over Rivals — ryanshrout · 2026-08-18
- Salvaging parts: building a local AI rig with 64GB RAM and a €1700 budget — joquinjack · 2026-08-18
- Mesh LLM: A Third Option Between Crypto Rigs and Cloud Subscriptions — alex_verem · 2026-08-18
- Overclocking VRAM on 4x RTX 5060Ti for LLM inference — Ok-Breakfast1878 · 2026-08-18