Home › News › AI Daily October 27 Afternoon Edition

AI Daily | October 27, 2026 Afternoon Edition: Anthropic Officially Files Its Record-Breaking IPO, OpenAI Halves GPT-6.1 Terra API Prices

Published: 2026-10-27 16:00  |  Category: AI Daily  |  Read: about 8 min

Top Stories This Afternoon

  • Anthropic officially files its S-1: Plans a Nasdaq listing under the ticker ANT with a targeted $2 trillion valuation, potentially the largest IPO ever
  • OpenAI launches GPT-6.1 Terra: API prices 50% lower than the previous generation, focused on visual reasoning and Computer Use
  • Moonshot AI's Kimi K3.1 Pro goes live: A 3T-parameter MoE architecture, the strongest open-weight model, with full weights promised by November 4
  • DeepSeek fully open-sources the DeepInfer inference framework: Compatible with NVIDIA and Huawei Ascend chips
  • CCTV reports an AI face-forgery fraud case: The Hangzhou Intermediate People's Court publishes the case and prevention tips

1. Anthropic Officially Files Its S-1, the Largest IPO in AI History

AnthropicIPONasdaq

Anthropic, the world's most valuable AI startup, officially filed its IPO registration statement (S-1) on October 26, planning to list on Nasdaq under the ticker "ANT". This will be the largest IPO in the history of the AI industry.

  • The targeted valuation is as high as $2 trillion, which would surpass the $1.77 trillion record set by SpaceX's listing in June this year
  • First-half 2026 revenue was $16.29 billion (about $11.5 billion in Q2 alone), up more than 13-fold year over year
  • But net loss reached $8.07 billion, underscoring the enormous investment in the AI compute race
  • Annualized revenue run rate (ARR) has passed $65 billion, with gross margin above 80%
  • Trading is expected to begin as early as mid-November, and NVIDIA plans to anchor the deal with up to $10 billion in stock
Editor's note: Anthropic's IPO is not only a capital-market spectacle but a milestone in the commercialization of the entire AI industry. Behind a 13-fold revenue increase in half a year is explosive enterprise AI demand, yet the $8.07 billion net loss also reminds investors that the AI compute race is still burning cash and the profitability inflection point has not arrived.

2. OpenAI Launches GPT-6.1 Terra, Cutting API Prices in Half

OpenAIGPT-6.1Terra

OpenAI has officially launched GPT-6.1 Terra, filling the mid-tier slot in the GPT-6 family. It focuses on visual reasoning, precise computer use and coding, while delivering a major price cut.

  • Standard API pricing: $1 per million input tokens, $5 for output and $0.05 for cached input
  • About 50%–58% cheaper than the previous generation GPT-5.6 Terra ($2/$12)
  • Only half the price of GPT-6.1 Sol and one-tenth of GPT-6 Astra
  • A context window of 1.05 million tokens and maximum output of 128,000 tokens
  • On the Artificial Analysis intelligence index, GPT-6.1 Terra (Max) scores 48, surpassing every Chinese open-weight model except Kimi K3.1 Pro
Technical breakthrough: GPT-6.1 Terra marks a shift in large-model competition from "who is stronger" to "who is more cost-effective." Halving API prices while improving performance means the value-for-money inflection point for AI applications has arrived, and the "working model" of the agent era has officially debuted.

3. Moonshot AI Releases Kimi K3.1 Pro, the Strongest 3T-Parameter Open Model

Moonshot AIKimiOpen Source

Moonshot AI has officially released Kimi K3.1 Pro, the first model in the K3.1 series. The company calls it the strongest open-weight model to date, focused on long coding tasks, 3D modeling and financial scenarios.

  • 3 trillion total parameters (MoE architecture) with a 1 million token context window
  • Ranks 4th globally and 1st among open-source models on the Artificial Analysis intelligence index
  • Terminal-Bench 4.0 score of 65.7%, an 11-point gain over the previous Kimi K3
  • A high-speed K3.1 Highspeed edition is released alongside it
  • Committed to releasing full weights by November 4
Open-source milestone: Kimi K3.1 Pro's release narrows the gap between open and closed models from 12 points to 3. Moonshot AI's commitment to fully open-sourcing the weights shows the determination and strength of Chinese large models in the open-source space.

4. DeepSeek Fully Open-Sources the DeepInfer Inference Framework

DeepSeekOpen SourceInference Framework

DeepSeek has fully open-sourced the DeepInfer inference framework, achieving compatibility with multiple chip platforms.

  • Compatible with NVIDIA and Huawei Ascend chips
  • Lowers the barrier to deploying large-model inference
  • Supports multiple hardware architectures, improving inference efficiency
  • Continues DeepSeek's consistent open-source strategy

5. CCTV Reports an AI Face-Forgery Fraud Case and Publishes Prevention Tips

AI SecurityCCTVDeepfake

According to CCTV, the Hangzhou Intermediate People's Court published a case in which criminals used AI to forge faces for fraudulent payments. The case reveals the risks of deepfake technology being misused.

  • Criminals used online photos to generate AI-forged faces
  • They bypassed facial recognition verification on some platforms
  • The court published prevention tips, reminding the public to better protect personal information
  • It called on platforms to upgrade facial recognition technology to counter deepfakes
Security warning: AI deepfake technology is increasingly being used for financial fraud. The public should stay alert, and platforms must keep upgrading biometric verification to counter new attack methods.

6. Chery Unveils the CheryMind Vertical Model and Ziwei Chips

CheryPhysical AIAutomotive

Chery Automobile has released CheryMind, a vertical foundation model for physical AI, along with the Ziwei chip series.

  • A vertical large model optimized specifically for automotive scenarios
  • Paired with in-house Ziwei chips for software-hardware synergy
  • Targets intelligent driving, in-vehicle interaction and similar scenarios
  • Marks automakers' accelerating push into core AI technology

7. World's First General Marine Foundation Model Clears Filing

Marine AIFilingShihang Intelligence

The general marine foundation model "CEORION", independently developed by Shihang Intelligent, has officially completed filing with the Cyberspace Administration of China, becoming the world's first general marine AI foundation model to clear filing.

  • The world's first general foundation model for the marine domain
  • Formally approved through Cyberspace Administration of China filing review
  • Marks China's leading position in marine AI globally
  • Will serve marine research, resource development and other scenarios

Today's Data at a Glance

MetricFigure
Anthropic targeted valuation$2 trillion
Anthropic first-half revenue$16.29 billion
Anthropic net loss$8.07 billion
Anthropic ARR$65 billion
GPT-6.1 Terra API price cut50%–58%
GPT-6.1 Terra context window1.05 million tokens
Kimi K3.1 Pro parameters3 trillion (MoE)
Kimi K3.1 Pro context1 million tokens
Kimi K3.1 Pro open-source commitmentBy November 4
DeepSeek DeepInfer compatible chipsNVIDIA + Huawei Ascend

Sources

  • CSDN: AI Industry Daily, October 27, 2026
  • Sina Finance: The largest IPO ever turns out to be the AI company that loses the most money
  • UC Browser: GPT-6.1 Terra launches at a rock-bottom price as the working model of the agent era arrives
  • NetEase News: Kimi releases K3.1 Pro, more capability at a lower price to challenge Opus 5.5
  • Toutiao Tech: AI Tech, October 27, 2026
  • International Finance News: The world's first general marine foundation model "Shihang CEORION" clears filing
  • Xueqiu: DeepSeek tops global weekly large-model call volume