⌘K
Change language Switch ThemeSign In
Narrow Mode
Predicting Token Price Trends Based on Wright's Law ===================================================
Predicting Token Price Trends Based on Wright's Law ===================================================  ### 李继刚
@lijigang
如果因资源所限（电力、显卡、内存等），阶段时间内模型输出的「智能」总量是有「上限」的，那么，Token Price 的走势会不会是：
- SOTA Token Price: 持平甚至涨价
- Normal Token Price: 指数下降
Mar 17, 2026, 2:58 AM View on X
1 Replies
2 Retweets
10 Likes
3,045 Views  李继刚 @lijigang
One Sentence Summary
The author analyzes the divergent price evolution paths of SOTA versus normal model tokens within the context of resource constraints.
Summary
This tweet analyzes AI model pricing trends by introducing Wright's Law. The author posits that due to resource limitations (power, GPUs, memory), the supply of state-of-the-art (SOTA) intelligence has a ceiling, leading to flat or even rising prices. Conversely, 'Normal' intelligence follows the law where unit costs drop by a fixed percentage as cumulative production doubles, resulting in exponential price decreases. This alternating pattern of 'SOTA with the left foot, Normal with the right' is driving the industry toward AGI.
AI Score
82
Influence Score 4
Published At Today
Language
Chinese
Tags
Token Price
Wright's Law
AGI
AI Economics
SOTA HomeArticlesPodcastsVideosTweets
Predicting Token Price Trends Based on Wright's Law | Bes... ===============