Think of AI tokens as the currency for running AI tasks, like sending messages or writing code.
They aren't cheap: Some models — for example Anthropic's Claude 4.6 Sonnet — have output-token rates around $0.000015 ( $15 per million output tokens); pricing varies across providers and between input and output tokens, and big projects can burn through them fast (Large projects can consume tokens rapidly, leading to substantial costs).
At a recent event, Huang said engineers could be given token budgets equal to roughly half their salary, which he said could help amplify their productivity by about 10x.
Now, job seekers are even asking about token or compute budgets during interviews.
It looks like token budgets might become the next big thing in Silicon Valley job offers.
Contact to : xlf550402@gmail.com
Copyright © boyuanhulian 2020 - 2023. All Right Reserved.