Test Shows 12 Extra Minutes of Model Thinking Costs 25x for 0.06 Point Gain

A blind test across 8 product management tasks found Fable 5.2's max-effort mode cost about 25x more (roughly $3.14 extra) for 12 minutes of thinking, yet scored only 0.06 points higher than a 30-second medium-effort run.

2026-11-11 ~ 2026-11-11 · 2 related posts

• Paying 25x for 0.06 points: blind test shows max reasoning effort barely beats medium — aakashgupta · 2026-11-11
• Fable 5.2 tested: max effort costs 25x more than medium for just +0.06 score — aakashgupta · 2026-11-11