GPT-6 Money Burn

Pick a wallet — $5, $10, $20, $50, $100 or $1000 — and hand an identical stack of cash to ten reasoning configurations: GPT-6 Astra at Max, xhigh, high and medium, plus GPT-5.6 Sol, Terra and Luna at their top tiers. Each gets its own campfire: a ring of stones, crossed kindling, and a strapped wad of notes burning on top. Particle flames rise from the burning surface and lick around the sides, the paper chars along a ragged ember line that eats inward, ash flakes peel off and tumble upward with the sparks, and the pile sinks into a mound of ash while smoke drifts off a dead fire. The hungrier the reasoning tier, the taller and brighter the flame. Every lane's timer freezes at the exact instant its balance hits zero, then an autopsy ranks who lasted longest with $/s, $/hour and $/8h day. All ten campfires fit on a phone screen at once — no scrolling. Token prices are OpenAI's published Standard-tier list rates (Astra $10/$50, Sol $4/$20, Terra $2/$12, Luna $0.20/$1.20 per million input/output tokens, checked 10 Sep 2026); reasoning tokens bill as output tokens, and effort changes how many you spend rather than the price per token. Decode throughput and the per-effort multiplier are not published by OpenAI, so those stay clearly-labelled estimates you can edit — change any cell and every flame, timer and ranking re-scales instantly from the same formula.