Vibeleaderboard
← All Intel
Intel / post

What a 10x price cut did to GPT-5.6 token demand

Source
x.com
Date
OpenRouter@OpenRouter
Thread · 5 parts

What happened when GPT 5.6 Terra and Luna were heavily discounted on OpenRouter? Token usage exploded by 13.8x Jevons Paradox = as technology makes the use of a resource more efficient, total consumption of that resource increases rather than decreases 🧵

Some of the additional tokens were from older OpenAI models. But the bulk of the increase were net new tokens + stealing share from competitors.

Importantly, many of the users with Terra/Luna usage during the discount window kept using the models after the discounts were retired. Overall tokens post-discount were actually higher on a daily average basis than during the window (mostly thanks to a few whales)

More insights on the impact of these discounts in the full piece: https://t.co/9mjQz0zBgu

Read the full thread on X

Context

Heavy discounts on OpenAI's GPT-5.6 models coincided with sharply higher usage on OpenRouter. Its retrospective reports that average daily volume during the July 27 to August 14 program rose 13.8 times for Luna and 5.6 times for Terra, each compared with the July 8 to 26 period before it. The 13.8 figure therefore belongs to Luna alone, not to both models together as the thread's headline could suggest. The earlier pricing announcement covers the rates, not these results. This is an observational company analysis and does not by itself prove the discounts caused the rise.

The company also reports that about 32% of the customers who used the models during the program used them again in the roughly six days after it ended, counted by customer rather than by volume. Daily average volume after the program was higher than during it, which the thread attributes mostly to a few very large accounts. That separates two outcomes: a minority of customers returned, yet their traffic sustained substantial volume. The follow-up window is short and may be incomplete.

Terms in this piece · Glossary
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
More from OpenRouter
Recommended reads
Comments

Checking sign-in…

Loading comments…