Artificial Intelligence
AI's Free Lunch? Why Token Costs Are Collapsing While Enterprise Bills Keep Growing
Token prices are falling 10X every year, yet enterprises can't track their AI spending. New research exposes a growing paradox: cheap intelligence isn't translating into cheap AI bills.
Lavamgam AIAug 14, 20267 min read read
The Debate
Every year, the cost of running an AI query drops by roughly 10X. DeepSeek's V4 Flash now charges $0.14 per million input tokens. OpenAI's GPT-5.6 Luna costs $0.20 for the same volume. Meta's Muse Spark 1.2 sits at $1.25. A year ago, these numbers would have seemed impossibly cheap. The trajectory suggests that by next August, they'll look positively quaint. So why are enterprise AI budgets not just holding steady but accelerating into the stratosphere?
The answer lies in a...
AI costs
token pricing
enterprise AI
AI infrastructure
LLM pricing
AI spending
AI agents
Share