Tag: #token-usage16 results
China's daily AI Token calls reached over 140 trillion in March 2026, up over 1,000x in two years from 100 billion in early 2024. National Data Bureau designates 2026 as 'data value release year' with six actions to boost AI datasets. Plans advance green computing synergy and national integrated computing network.
DeepSeek V4 Flash leads OpenRouter’s weekly global token-usage ranking with 7.1 trillion tokens. Four of the top five models are now Chinese, reportedly prompting OpenAI to reduce GPT-5.6 Luna pricing by 80%.
Tencent Hunyuan Hy3 preview, online for two weeks, has surpassed previous Hy2 model's total Token calls by 10x. Usage surged notably in code and agent scenarios. Apps like WorkBuddy, CodeBuddy, and Qclaw saw over 16.5x growth.
Disney tracks employee Claude usage via an AI dashboard, with one employee making 460K calls in 9 days. Meta burned 60 trillion tokens in a month across 85K employees, costing ~$9B equivalent. Widespread adoption spans lawyers, educators, Hollywood, and finance beyond just coding.
ByteDance's Doubao large model achieved over 120 trillion daily Token usage. Broadcasting industry issued strict statement against AI face-swapping. Other updates include auto industry leadership changes and L2 ADAS advancements.
OpenRouter data shows Chinese AI large models reached 4.69 trillion weekly Tokens as of March 15, surpassing the US for the second week. They claimed the top three global invocation spots. JPMorgan predicts China's AI inference Token use will grow 370x from 2025 to 2030.
Lobster users are voraciously consuming domestic models, hitting 4.19 trillion token calls—a 34.9% surge surpassing the US. Domestic models claim the top three spots in 龙虾 applications. This underscores rising dominance of Chinese LLMs at home.
Companies are increasingly implementing controls to prevent employees from overspending on AI tokens for minor tasks. The industry is shifting from an era of unchecked AI experimentation to one of strict token rationing.
Lingzhu AI's daily token consumption has surpassed 5 billion. Its user base is far smaller than major LLM companies and platforms, yet per-user usage is exceptionally high. Deep engagement may form a key moat for AI applications.
GitHub fixed a token counting bug in Copilot that undercounted usage, disrupting the pricing model. The repair caused rapid exhaustion of subscription allowances due to accurate counting. Customers are protesting and urged to cut usage amid server strain.