AI Daily | October 20, 2026: Chinese LLM API Calls Top the World for 25 Consecutive Weeks

Published: 2026-10-20  |  Category: AI Daily  |  Read: about 6 min

Today's Key Stories

  • China's weekly LLM API calls exceed the US for the 25th consecutive week, with DeepSeek V4.1 Flash ranking first worldwide for two weeks in a row
  • OpenAI accelerates a $30 billion funding round, locking in capital ahead of Anthropic's IPO
  • Zhou Xiaochuan: AI-driven productivity gains come mainly from the application layer
  • High-quality data becomes AI's "hard currency" as LLM competition enters its second half
  • Anthropic and Nscale agree a $45 billion cloud computing partnership

1. Chinese LLM API Calls Lead the World for 25 Consecutive Weeks

DataDomestic Models

According to the latest data from OpenRouter, global AI LLM API calls totaled 164 trillion tokens last week (October 12 to 18), down 4.65% week over week. Chinese models accounted for 79.18 trillion tokens, and the top five models by global call volume were all Chinese generative LLMs.

DeepSeek V4.1 Flash topped global call volume for two consecutive weeks, and other Chinese models such as StepFun also performed strongly. The data marks more than half a year in which China has continuously led the US in the scale of LLM usage.

RankModelWeekly calls (trillion tokens)Trend
1DeepSeek V4.1 Flash~22.5No. 1 for two consecutive weeks
2StepFun Step 5~18.3Rising
3Qwen 2.5~15.7Stable
4Doubao~12.8Rising
5Kimi K2~9.8Stable
Trend analysis: The explosive growth of Chinese models at the application level comes down to extremely low pricing and a rich API ecosystem. At just $0.25 per million tokens, DeepSeek V4.1 Flash has sharply lowered the barrier to entry for developers and enterprises.

2. OpenAI Accelerates $30 Billion Funding Round

FundingOpenAI

According to reports, OpenAI is stepping up efforts to secure $30 billion in funding from investors, aiming to lock in the money before Anthropic launches its IPO. SoftBank's earlier $30 billion investment commitment was fully funded in early October, and OpenAI's annualized revenue is approaching $50 billion, with a pre-money valuation of about $1.4 trillion.

Meanwhile, Anthropic is seeking to launch its IPO as early as mid-November, with its valuation possibly cut by more than $500 billion from the previously expected $2 trillion. The capital race between the two AI giants is entering a white-hot phase.

Market reaction: After the news broke, Nvidia's market value evaporated by $168.4 billion overnight, raising doubts about the sustainability of the AI investment boom. Still, most analysts believe revenue growth at leading AI companies is still accelerating and that valuation-bubble risk remains manageable.

3. Zhou Xiaochuan: AI-Driven Productivity Gains Come Mainly from the Application Layer

OpinionPolicy

On October 20, at the 2026 Bund Annual Conference, co-hosted by the China Finance 40 Forum (CF40) and Tsinghua University, Zhou Xiaochuan, former governor of the People's Bank of China, said that the speed and scale of AI-driven productivity gains in China will be reflected mainly in the application layer rather than the foundation-model layer.

Zhou pointed to two key paths:

  1. Model distillation: compress and accelerate large models so their capabilities can be pushed down to edge devices
  2. Application innovation: draw on China's rich industrial scenarios to put AI to work in manufacturing, finance, healthcare and other sectors
Editor's note: Zhou Xiaochuan's view hits the mark. China still trails the US in foundation-model R&D, but it has overtaken it in the speed and scale of application-layer innovation. Weekly call volume of 164 trillion tokens is the best proof.

4. High-Quality Data Becomes AI's "Hard Currency"

TrendData

At the 2026 Sustainable Global Leaders Summit held in Shanghai from October 20 to 22, several Nobel and Turing Award laureates noted that LLM competition has entered its second half and that high-quality data is becoming AI's "hard currency".

The "2026 International AI Safety Report" notes that frontier AI development is driven by three factors: compute, algorithmic progress and data. As public text resources run dry, data has become the most critical bottleneck.

Experts warn: "Future development may be limited by data availability." Compliance and precision will drive the next stage of model competition.

5. Anthropic and Nscale Reach a $45 Billion Partnership

PartnershipInfrastructure

UK AI cloud company Nscale and Anthropic have agreed a $45 billion cloud computing partnership, one of the largest infrastructure deals in AI to date.

In addition, AI chip startup Tenstorrent closed a new $1.5 billion funding round, further enriching the AI compute ecosystem.

6. Other Key Developments

ProductsOpen Source

  • MiniMax M3.1 Flash released: supports a 1-million-token context, with input priced at just $0.14 per million tokens, setting a new low
  • Amazon TorchNeuron: announced at PyTorch Conference North America, enabling PyTorch models to run on its in-house Trainium chips, with an official open-source release planned for October 20
  • China Telecom's Chenxing general AI lab open-sourced the Xing4.0-29B-Intent model
  • VAST EdenLive: achieves real-time AI video generation at 720p@24fps on an RTX 5090
  • NVIDIA won a CPU order from AI startup Modal (which had previously relied entirely on Intel and AMD)

7. Today's Data at a Glance

MetricFigureChange
Global weekly LLM API calls164 trillion tokens-4.65% week over week
Share of Chinese models79.18 trillion tokens (48.3%)Rising steadily
OpenAI valuationAbout $1.4 trillion—
Anthropic IPO expectationAs early as mid-November—
MiniMax M3.1 Flash price$0.14 per million input tokensLowest in the industry
Nscale-Anthropic partnership$45 billion—

What to Watch Tomorrow

  • The 2026 Bund Annual Conference continues, with more AI policy signals expected
  • Details of the official Amazon TorchNeuron launch
  • Tracking progress of the Anthropic IPO

— This article is compiled and published daily by AICat —

Sources

  • National Business Daily, "China's Weekly LLM API Calls Exceed the US for 25 Consecutive Weeks", 2026-10-20
  • Sina Finance, "Zhou Xiaochuan: AI-Driven Productivity Gains Come Mainly from the Application Layer", 2026-10-20
  • Sina Finance, "LLM Competition Enters Its Second Half: High-Quality Data Becomes AI's Hard Currency", 2026-10-20
  • SegmentFault, "AI Daily (October 20, 2026)", 2026-10-20
  • Baidu Baijiahao, "Tech News Roundup for October 20, 2026", 2026-10-20
← Back to News