Home › News › AI Daily October 27 Evening Edition

AI Daily | October 27, 2026 Evening Edition: NVIDIA and Meta Deepen AI Agent Collaboration, Anthropic's Record IPO Sets the Market Alight

Published: 2026-10-27 20:00  |  Category: AI Daily  |  Read: about 7 min

Top Stories This Evening

  • NVIDIA and Meta deepen collaboration: Expanding their Muse-series model partnership to drive AI agent innovation and deploying VeraRubin clusters
  • Anthropic files its IPO prospectus: Targeting a valuation of up to $2 trillion with $16.29 billion in first-half revenue, the largest AI company IPO ever
  • CCTV reports an AI face-forgery fraud case: The Hangzhou Intermediate People's Court publishes a case in which criminals used AI face-swapping to commit financial fraud
  • Chery unveils the CheryMind model: A vertical foundation model for physical AI, launched with the Ziwei chip series
  • DeepSeek retains global No. 1 in weekly call volume: Chinese large-model weekly call volume reaches 93.54 trillion tokens, up 18.14% month on month

1. NVIDIA and Meta Deepen Collaboration to Drive AI Agent Innovation

NVIDIAMetaAI Agents

On October 27, 2026, NVIDIA and Meta announced that they will further expand their collaboration on Muse-series models to meet growing market demand for AI agents. The move marks a new level of strategic synergy between the two companies in AI.

  • The two sides will use OpenShell and Sentry technology to strengthen AI agent security
  • Expand VeraRubin cluster deployments to improve model training and inference performance
  • The collaboration focuses on the reliability and controllability of AI agents
  • It is another deep collaboration in frontier models following Muse Wonder
Editor's note: The deepened NVIDIA–Meta collaboration sends a clear signal: AI agents are moving from the lab into production environments. When a compute giant and a social platform join forces to solve agent safety issues, it means the infrastructure for large-scale commercial deployment is maturing fast.

2. Anthropic Files Its IPO Prospectus, the Largest AI Company IPO Ever

AnthropicIPONasdaq

Anthropic, the world's most valuable AI startup, officially submitted its S-1 prospectus to the US SEC on October 26, planning to list on Nasdaq under the ticker "ANT" with a targeted valuation of up to $2 trillion, potentially the largest tech IPO in history.

  • The prospectus runs 335 pages, detailing its financial position and compute commitments
  • First-half 2026 revenue was $16.286 billion, up about 13-fold year over year
  • Net revenue retention (NRR) of about 900%, showing exceptionally strong customer stickiness
  • Compute commitments: Microsoft $44.9 billion, Google $114.5 billion, SpaceX $120.5 billion
  • The founding team holds 50.1% of the shares, retaining control of the company
  • Plans to list officially in mid-November
Risk note: Despite stunning revenue growth, Anthropic's huge losses are equally noteworthy. The prospectus shows the company's first-half net loss reached tens of billions of dollars, mainly due to large-scale investment in compute infrastructure. It is a high-stakes bet that trades losses for the future.

3. CCTV Reports an AI Face-Forgery Fraud Case, Sounding the AI Security Alarm

AI SecurityCCTVDeepfake

According to CCTV, the Hangzhou Intermediate People's Court published a typical case in which criminals used AI face-forgery technology to commit fraudulent financial payments, drawing widespread public attention. The court also published prevention tips.

  • Criminals obtained victims' photos online and used AI face-swapping to bypass facial recognition
  • The case involved the payment systems of multiple financial platforms
  • The court advises against uploading high-resolution frontal photos to public platforms
  • It recommends that financial platforms add multimodal biometric verification
Security warning: AI deepfake technology is evolving from a technical gimmick into a real threat. As AI face-swapping becomes cheaper, personal biometric information security has become a new line of defense in the digital age. Tech companies and regulators need to accelerate the creation of more robust verification mechanisms.

4. Chery Unveils the CheryMind Model as Automotive AI Enters the Physical Intelligence Era

CheryCheryMindPhysical AI

Chery Automobile has officially released CheryMind, a vertical foundation model for physical AI, along with the accompanying Ziwei chip series, marking an important step by Chinese automakers into AI large models.

  • CheryMind is a vertical large model designed for automotive scenarios
  • The Ziwei chip series provides on-device AI inference capability
  • It covers intelligent driving, cabin interaction, vehicle diagnostics and more
  • It reflects the industrial convergence trend of AI plus manufacturing

5. DeepSeek Retains Global No. 1 in Weekly Call Volume

DeepSeekGlobal Call VolumeOpenRouter

According to the latest OpenRouter data, as of October 27, 2026, weekly call volume for Chinese AI large models reached 93.54 trillion tokens, up 18.14% month on month. Among them, DeepSeek retained the global No. 1 spot in weekly call volume.

  • Chinese large models have led global call volume rankings for more than 26 consecutive weeks
  • DeepSeek's open-source strategy and high cost-performance are the core reasons for its continued lead
  • Hong Kong-listed large-model concept stocks rose on the news, with Zhipu up more than 10%
  • Global AI developers' reliance on Chinese models continues to deepen
Data highlight: Weekly call volume of 93.54 trillion tokens means Chinese large models' influence in the global AI ecosystem is still expanding fast. The open-source plus low-price strategy is reshaping competition in the global AI market.

6. GPT-6.1 Terra Keeps Cutting Costs as API Costs Fall Another 27.7%

OpenAIGPT-6.1Price Cuts

OpenAI's GPT-6.1 Terra continues to push cost optimization, with average API cost down another 27.7%. Data shows KV reads account for 82.4% of API cost, and CMV lossless session pruning has become the key to reducing costs.

  • Sharply lowers usage costs while maintaining intelligence levels
  • KV cache optimization is the core technical lever for cost reduction
  • The large-model API price war intensifies further
  • Developers will get better value for money

7. Kimi K3.1 Pro Open-Source Countdown as Moonshot AI Promises a November 4 Release

Moonshot AIKimiOpen Source

Kimi K3.1 Pro from Moonshot AI is now live, with a promise to be officially open-sourced on November 4. This makes it the second Chinese AI company after DeepSeek to open up its core model weights.

  • Kimi K3.1 Pro performs strongly across multiple benchmarks
  • The open-source commitment reflects Chinese AI companies' openness
  • Open-sourcing will accelerate community ecosystem building and application innovation
  • It complements DeepSeek's open-source strategy

Today's Data at a Glance

MetricFigure
Anthropic targeted valuationUp to $2 trillion
Anthropic first-half revenue$16.286 billion
Anthropic revenue YoY growthAbout 13-fold
Anthropic NRRAbout 900%
Anthropic prospectus length335 pages
Chinese large-model weekly call volume93.54 trillion tokens
Weekly call volume MoM growth18.14%
GPT-6.1 Terra API cost reduction27.7%
KV reads share of API cost82.4%
Chinese large models' consecutive leadMore than 26 weeks

Sources

  • Sohu Finance: NVIDIA and Meta deepen collaboration to drive AI agent innovation
  • Sina Finance: The largest IPO ever turns out to be the AI company that loses the most money
  • CSDN: AI Industry Daily, October 27, 2026
  • Toutiao: Today's AI News, 2026-10-27
  • Toutiao: AI Tech, October 27, 2026
  • Xueqiu: DeepSeek tops global weekly large-model call volume
  • Sina AI Hourly Hotspots: October 27, 2026