Gate News message, April 20 — According to OpenRouter data, global AI model call volumes totaled 206 trillion tokens last week (April 13-19), marking a second consecutive week of decline. China’s AI models saw weekly call volumes drop to 444.1 trillion tokens, down 23.77% from the previous week, while U.S. AI models reached 490.8 trillion tokens, up 20.62% week-over-week. This marks the first time in two months that U.S. AI model call volumes have surpassed China’s.
Among the top nine global AI models, four are Chinese. DeepSeek V3.2 ranked second with 1.28 trillion tokens in weekly calls. MiMo-V2-Pro ranked fourth with 1.15 trillion tokens, up 90% week-over-week. MiniMax had two models in the top nine: MiniMax M2.5 ranked sixth with 1.05 trillion tokens, while MiniMax M2.7 ranked seventh with 961 billion tokens, down 19% from the previous week.
Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to
Disclaimer.
Related Articles
Nvidia CEO Huang Renxun Refutes AI Doomsday Claims: The Future Will Create Large-Scale Emerging Employment Opportunities
NVIDIA CEO Jensen Huang said at a global panel discussion that AI is not an apocalyptic scenario, and he believes AI will drive U.S. reindustrialization and create jobs. He predicted that the next wave of “agentic AI” will cause GPU demand to surge by a thousand times, requiring new types of industrial-grade infrastructure and a massive labor market. He emphasized that AI increases productivity rather than replacing human workers, and he called for responsible development and cross-border safety standards. The impacts on employment and the economy still need to be monitored over the long term.
ChainNewsAbmedia20m ago
OpenAI Accelerates AI Agent Phone to H1 2027, MediaTek to Secure Exclusive Processor Deal
According to analyst Ming-Chi Kuo's latest industry survey, OpenAI is accelerating development of its first AI agent phone and aims for mass production as early as H1 2027. MediaTek is now more likely to secure an exclusive processor order, with the chip based on a customized version of
GateNews23m ago
Sierra raises $950 million, valuation $15.8 billion: Bret Taylor secures the AI customer service leader with an OpenAI chairman role
Sierra, co-founded by Bret Taylor, announced it has completed a $95 million Series E with a valuation of $15.8 billion, led by Tiger Global and GV. In eight quarters, it reached $150 million ARR, with a Fortune 50 penetration rate exceeding 40%. It is positioned as a vertical enterprise customer service platform centered on AI agents, offering plug-and-play workflows. Taylor also serves as Chair of OpenAI, with governance and interest disclosures in focus.
ChainNewsAbmedia2h ago
Gemini API adds Webhooks: Google solves the pain point of long-task polling, and Batch/Veo can be pushed instantly
Google Gemini API launched Webhooks on May 4, automatically pushing results to the callback URL when long-running tasks complete, replacing polling to reduce resource usage and latency while simplifying code. It applies to Batch API, Veo2, and long-context inference, and is especially suitable for serverless. Compared with OpenAI leaning toward SSE and Anthropic still using polling, Google emphasizes developer infrastructure. Going forward, attention will focus on security mechanisms and scaling models. For Taiwan developers, integrating it now can significantly reduce quotas and system load.
ChainNewsAbmedia2h ago
Krutrim Shifts to AI Cloud Services on May 5, Posts First Profit as FY26 Revenue Hits $31.6M
According to Press Trust of India, Indian AI firm Krutrim repositioned itself as a domestic AI cloud services provider on May 5, pausing its chip design efforts and redirecting capital and talent to cloud infrastructure.
The company reported FY26 revenue of approximately 3 billion rupees (US$31.6 m
GateNews3h ago
