Tokenmaxxing is back, but only if it's affordable.
AI inference costs dropped 43% in ten weeks to $1.16 per million tokens as OpenAI slashed GPT-5.6 prices 80% and Chinese models intensify