AI News
Nov 12, 2026 · WednesdayPublic sources · Daily newsPublished daily at 08:00 Beijing time

AI News Daily

AI developments selected from public sources

Issue 20412Nov 2026Wednesday
11monthThis month 12 issues
12
major stories
28
sources
1
first-party stories
About 7 min read

Lead story

Comparison of Gemini Decision and other APIs against various benchmarks

Google releases Gemini Decisions API and Gemini Decision 1 model

Google launches the Gemini Decisions API, powered by the new model Gemini Decision 1 (gemini-decision-1), which can evaluate up to 1,000 yes/no, multiple-choice, and rating questions on text, image, audio, video, and PDF inputs in a single forward pass, returning confidence scores from 0 to 1. The model has an end-to-end latency of 150-250 ms and is priced at $0.10 per 1 million input tokens, with output tokens not billed; it can be used by obtaining an API key in Google AI Studio.

First-party sourcesGoogle BlogAlso 4 other sources reported thisOriginal ↗
01

Model releases / updates

Baidu releases ERNIE 5.5 Flash Preview, focused on end-to-end coding, and announces it will soon be open-sourced

Baidu has released ERNIE 5.5 Flash Preview, with 280 billion total parameters and 14 billion activated per inference, support for a 1 million token context, and a generation speed exceeding 200 tokens per second, and announced that it will soon be open-sourced. The model focuses on strengthening coding and agent capabilities, and the company says it can compete with Claude Opus 5.5 and GPT-6.1 Sol on benchmarks such as Terminal-Bench and AutomationBench. The author tested it by orchestrating animations with Python code, generating a playable Backrooms web game, and building a Suzhou garden in Blender. The weights are expected to be released soon under Apache 2.0 with support for SGLang and vLLM deployment; the API is already live on Baidu AI Cloud's Qianfan platform, with a limited-time 50% discount of 1 yuan per million input tokens and 4 yuan per million output tokens.

Original

Alibaba Qwen4-Max preview launches, limited-time 10% discount on token prices

The Alibaba Qwen4-Max preview has launched ahead of its official release and will offer a 10% discount on token prices for a limited time. The new model will be integrated into Qoder (for coding), Qwen Office and QwenStudio, and the company says it is stronger in knowledge, world understanding, and coding and agent capabilities. Qwen4-Max is currently available for early access on Alibaba Token Plan as well as QwenWork, Qoder, Qwen APP and QwenStudio.

Original
02

Product release/Updated

AWS releases AWS Emulator, a local simulator written by autonomous AI agents

On November 11, 2026, AWS released AWS Emulator, a local simulator built on 438 AWS API models and covering more than 200 services, allowing developers and coding agents to call AWS APIs on their own machines without an AWS account. Matthew Goldberg, an AWS senior principal engineer, said the team neither wrote nor reviewed the simulator's code; humans handled specifications, testing mechanisms, merge gates, and the release pipeline, while three agent teams were responsible for writing tests, locating problems, and fixing them, resolving thousands of issues per day. The code is released under CC0 (vendor code under Apache 2.0), the repository does not accept PRs, and defects can be reported via GitHub Issues; the simulator does not evaluate IAM policies and is not suitable for production workloads or security testing.

Original

Anthropic releases three AI office tools: Claude Sheets, Whiteboards, and Data

Anthropic has released three AI office tools, Claude Sheets, Claude Whiteboards, and Claude Data, with the new features rolling out on web, desktop, and mobile starting the same day. Claude Sheets supports generating and editing spreadsheets with formulas and tabs together with Claude, Claude Whiteboards can turn brainstorming content into documents or slides, and Claude Data provides authenticated queries and real-time sharing, with authenticated queries and real-time sharing still in testing. Anthropic also upgraded interactive connectors, allowing partners to embed apps such as Adobe and Atlassian directly into Claude.

Original

OpenAI launches ChatGPT Business Hub enterprise information pages and Business Agents

OpenAI has launched the Business Hub enterprise information page feature in ChatGPT, allowing businesses to claim and verify their business profiles, manage their name, address, business hours, contact information and other details, and obtain a blue verification badge through verification. Retailers can upload product data feeds containing prices and inventory, and eligible Shopify merchants can connect their stores through Business Hub or the Shopify admin and sync products to Shopify Catalog. OpenAI has also launched Business Agents, which allow businesses to create AI assistants trained on their own product and policy information; in the future these can also serve as promotional entry points in ChatGPT's advertising system, with paid promotional content clearly labeled. Businesses in ChatGPT is open for testing to eligible businesses in available regions, while Business Agents is currently limited to eligible e-commerce and online service businesses in the United States.

Original

Huawei Kirin 9050 series features next-generation Da Vinci architecture NPU, supports on-device running of 30B large models

Huawei Kirin officially detailed the on-device AI capabilities of the Kirin 9050 series, which features a next-generation Da Vinci architecture NPU and can support a 30B MoE large model running offline on a phone, completing complex task planning, intelligent photo editing and multimodal perception. The NPU performance of the Kirin 9050 Pro is 140% higher than the previous generation with 66% lower power consumption, and it integrates a Transformer engine, an ultra-low-power perception engine and the CANN heterogeneous computing architecture. The series also supports a 6B image editing model, a 7B Dense text model and small multimodal perception models; the 30B MoE model has 30 billion total parameters and about 2 billion activated parameters, with additional memory usage of less than 2.5GB.

Original
03

Industry news

Anthropic opens IPO subscriptions, seeking to raise up to about $90.4 billion

Anthropic plans to issue 825 million shares at $70 to $78 per share, alongside a concurrent $26 billion private placement, for combined fundraising of up to about $90.4 billion, which could surpass SpaceX's planned IPO to become the largest stock offering in the world. The company is seeking a fully diluted valuation of about $1.4 trillion, with Goldman Sachs and Morgan Stanley leading the underwriting and 20 other banks participating, and plans to list on the Nasdaq Global Select Market under the ticker ANT. Some existing investors are reselling their IPO subscription allocations or selling indirect interests through SPVs, raising market concerns about selling pressure in the early days of trading.

Original

OpenAI third-quarter revenue reaches $10 billion, with quarter-over-quarter growth rising to nearly 50%

The Information, citing people familiar with the matter, reported that OpenAI disclosed at an investor call that third-quarter revenue reached $10 billion, 2.6 times the same period last year, with quarter-over-quarter growth rising from 17% in the second quarter to nearly 50%. Operating losses in the same period widened from $10.1 billion to $10.9 billion, while the adjusted operating margin improved from -150% to -109%. OpenAI is seeking to raise a new round of funding at a valuation of $1.25 trillion, after having lowered its valuation from $1.4 trillion and reduced its fundraising target to up to $20 billion.

Original

Report says Jensen Huang-linked matter led to $7.5 billion of Nvidia chips flowing into China

According to The Information, citing two people familiar with the matter, Alice Huang, who is connected to Nvidia CEO Jensen Huang, moved at least $7.5 billion worth of Nvidia chips into China through a corporate structure, involving more than 240,000 A100, H100 and H200 systems. The routes involved Century Litong, a Chinese company established in mid-2023, and Megaspeed, set up in Singapore, which was added to Nvidia's preferred cloud partner list in October 2023. After the news emerged, Nvidia's market value shrank by more than $400 billion in a single day; Jensen Huang has acknowledged the matter and launched an internal investigation, and multiple U.S. agencies are urging the Commerce Department to blacklist Megaspeed.

Original

Report: Anthropic IPO several times oversubscribed, raising up to about $64.35 billion

According to media reports citing sources, Anthropic's IPO has been several times oversubscribed, with subscription orders reaching several times the number of shares available. Regulatory filings show the company plans to issue and sell a total of 825 million shares at $70 to $78 per share, for an IPO size of up to about $64.35 billion, and to raise an additional $26 billion through a private placement. Based on the upper end of the offering price, the basic market value after the IPO and private placement would be about $1.02 trillion, with a fully diluted valuation of up to about $1.4 trillion; the company expects to set the final offering price on November 18 and begin trading on the Nasdaq Global Select Market under the ticker ANT on November 19.

Original

Anthropic reportedly considering raising IPO offering price, target valuation could reach $1.5 trillion

According to media reports, one day after disclosing IPO pricing details, Anthropic's underwriters have signaled that, given strong investor demand, they are considering raising the offering price above the $70-$78 range published in Tuesday's prospectus, potentially pushing the fully diluted target valuation to $1.5 trillion or even higher, with total fundraising possibly exceeding $100 billion. Anthropic plans to issue and sell a total of 825 million shares, for an IPO size of up to about $64.35 billion at the upper end of the range, and Google, Nvidia and Microsoft will together invest $26 billion at the IPO price. The company expects to set the final offering price on November 18 and begin trading on the Nasdaq Global Select Market on November 19.

Original
04

Briefs

End of issue

AI News DailyAutomaticallycompiledfrom public reports, with source links · View archive