Skip to main content
ModelWool

AI Daily · Nov 12

Nov 11 06:00 – Nov 12 06:00 (Beijing time) · 29 items

Updated 11/12 06:19 (Beijing time)

Today's perks

New perks in this issue come first; if there are fewer than 3, we add perks you can still claim.

  1. NewOpenRouterBaidu ERNIE releases ERNIE 5.5 Flash Preview modelBaidu ERNIE has released ERNIE 5.5 Flash Preview, with 280B total parameters and 14B activated parameters, supporting a 1M context. The API is now live on the Qianfan platform, and the weights will be open-sourced by Nov 25; the model is offered as a free variant on OpenRouter.
  2. NewAnthropicAnthropic grants a free quota reset to Pro, Max, Ultra, and Team usersAnthropic announced a free full quota reset for Pro, Max, Ultra, and Team users, covering all models and limits, available until 2:00 PM Pacific Time on Dec 10. The company said it recently fixed several minor issues that caused quota consumption to exceed expectations.
  3. NewAnthropicAnthropic offers free quota reset for paid Claude usersAnthropic announced a free full quota reset for Claude Pro, Max, Ultra, and Team users, covering all models and usage limits, while Claude now supports more MCP Apps extensions.
  4. NewZhipu AIZhipu, MiniMax and other LLM vendors offer 50% off for Tmall Double 11On Nov 11, LLM vendors including Zhipu, MiniMax, Qwen Office, Alibaba Cloud, and Qoder launched 50% off promotions for Tmall Double 11.

Today's highlights

  1. Google AIGoogle DeepMind releases AlphaGenome 2 gene sequence modelGoogle DeepMind has released AlphaGenome 2, which can predict the impact of DNA variants on specific cell types and is nearly twice as effective at identifying pathogenic variants as the previous best model; the model was trained on single-cell atlases of more than 100 million cells across 68 tissue types and is now available via API for non-commercial research.
  2. Google AIGoogle releases Gemini Decision model and Decisions APIGoogle releases Gemini Decision (gemini-decision-1) and the Decisions API, which can quickly make calibrated binary, categorical, and rubric-based decisions on text, images, audio, video, and PDFs, with zero output tokens.
  3. Alibaba QwenAlibaba Qwen4 Launch Imminent, Qwen4-Max Preview Now on Token PlanAlibaba's new-generation Qwen4 large model is about to be released. The Qwen4-Max preview is now available first on Alibaba Token Plan, with a limited-time trial at as low as 90% off; the personal Lite/Essential plans start at a limited-time daily price of ¥39/month, and new customers can enjoy an additional 50% off during the promotion, as low as ¥19.5.

Perks and offers

  • OpenRouter

    Baidu ERNIE releases ERNIE 5.5 Flash Preview model

    Baidu ERNIE has released ERNIE 5.5 Flash Preview, with 280B total parameters and 14B activated parameters, supporting a 1M context. The API is now live on the Qianfan platform, and the weights will be open-sourced by Nov 25; the model is offered as a free variant on OpenRouter.

  • Anthropic

    Anthropic grants a free quota reset to Pro, Max, Ultra, and Team users

    Anthropic announced a free full quota reset for Pro, Max, Ultra, and Team users, covering all models and limits, available until 2:00 PM Pacific Time on Dec 10. The company said it recently fixed several minor issues that caused quota consumption to exceed expectations.

  • AnthropicThird-party report

    Anthropic offers free quota reset for paid Claude users

    Anthropic announced a free full quota reset for Claude Pro, Max, Ultra, and Team users, covering all models and usage limits, while Claude now supports more MCP Apps extensions.

  • Zhipu AIThird-party report

    Zhipu, MiniMax and other LLM vendors offer 50% off for Tmall Double 11

    On Nov 11, LLM vendors including Zhipu, MiniMax, Qwen Office, Alibaba Cloud, and Qoder launched 50% off promotions for Tmall Double 11.

Models and products

  • Google AI

    Google DeepMind releases AlphaGenome 2 gene sequence model

    Google DeepMind has released AlphaGenome 2, which can predict the impact of DNA variants on specific cell types and is nearly twice as effective at identifying pathogenic variants as the previous best model; the model was trained on single-cell atlases of more than 100 million cells across 68 tissue types and is now available via API for non-commercial research.

  • Google AI

    Google releases Gemini Decision model and Decisions API

    Google releases Gemini Decision (gemini-decision-1) and the Decisions API, which can quickly make calibrated binary, categorical, and rubric-based decisions on text, images, audio, video, and PDFs, with zero output tokens.

  • Alibaba Qwen

    Alibaba Qwen4 Launch Imminent, Qwen4-Max Preview Now on Token Plan

    Alibaba's new-generation Qwen4 large model is about to be released. The Qwen4-Max preview is now available first on Alibaba Token Plan, with a limited-time trial at as low as 90% off; the personal Lite/Essential plans start at a limited-time daily price of ¥39/month, and new customers can enjoy an additional 50% off during the promotion, as low as ¥19.5.

  • Google AI

    Gemini API Managed Agents now support self-hosted sandbox environments

    Google announced that Managed Agents in the Gemini API now support bringing your own sandbox and using self-hosted environments, and developers can start building in AI Studio today.

  • OpenAI

    ChatGPT desktop app launches Finances feature

    OpenAI has launched the Finances feature in the ChatGPT desktop app, helping users save money, understand their spending, and make sense of investments, and also build custom financial tools; Free, Go, Plus, and Pro users can use it after updating the desktop app.

  • Google AI

    Gemini launches on Apple CarPlay

    Gemini is now available on Apple CarPlay. Users can tap the Gemini app icon on the CarPlay screen or sidebar to continue conversations from their phone, explore topics, or brainstorm by voice with Gemini Live.

  • Google AI

    Google launches experimental AI tool Tabletop

    Google Labs has released Tabletop, an experimental AI tool that generates customized collaboration hubs for coordinating with family and friends. It supports building interactive itineraries, polls, and brackets, and a waitlist is now available.

  • Tencent Hunyuan

    Tencent Hunyuan Hy4 launches on OpenRouter

    Tencent Hunyuan officially announced on X that Hy4 is now available on OpenRouter.

  • xAI

    Musk promotes Grok @Bot: hire it as a colleague for the hardest tasks

    Musk posted on X to promote Grok @Bot, saying you can "hire" it as a colleague and hand it the hardest tasks.

  • Tencent Hunyuan

    Tencent Hunyuan Hy4 API now available on Atlas Cloud

    Tencent Hunyuan announced that the Hy4 API is now available on Atlas Cloud, allowing developers to access Hunyuan Hy4 model capabilities through the platform.

  • MiniMax

    MiniMax H3 Image supported by fal, runware and other platforms

    MiniMax has released H3 Image, claiming it generates 4K images in about 10 seconds, supports 16 reference inputs and delivers a 2–3x speed improvement, targeting advertising, e-commerce and scaled content creation; platforms including fal and runware already provide support.

  • Cursor

    Cursor launches multiplayer Projects, with agents collaborating across teammates

    Cursor announced that Projects has been rebuilt for multiplayer collaboration, where each teammate and their agents can work together in the same workspace, and agents can also coordinate with one another.

  • Mistral AI

    Mistral uses Voxtral Transcribe 3 to transcribe conference talks in real time

    At AI Engineer Paris, Mistral used Voxtral Transcribe 3 to transcribe all main stage talks in real time, improving accuracy by feeding speaker names and terminology through contextual biasing.

  • GitHub

    GitHub App supports enterprise Copilot seat management API

    GitHub announced that enterprise-level GitHub Apps can now call the enterprise-level Copilot seat management API to control which users and enterprise teams are assigned Copilot seats.

Media reports

  • Related companiesOpenAIThird-party report

    OpenAI to retire ChatGPT legacy voice modes on Dec 11, shifting to GPT-Live

    OpenAI announced that it will retire ChatGPT's Advanced Voice Mode and Standard Voice Mode on Dec 11, 2026, fully transitioning to GPT-Live, which uses a full-duplex architecture.

  • Related companiesTencent HunyuanThird-party report

    Tencent launches Yuanbao Search: connects 157 authoritative sources, clearly separates ads from answers

    On Nov 11, Tencent's 28th birthday, the new search product Yuanbao Search officially launched. It is accessible at so.qq.com or by upgrading QQ Browser, connects 157 authoritative sources, can search Official Accounts and Channels, clearly separates ads from answers, and supports collapsing ads.

  • Related companiesOpenAIAnthropicThird-party report

    New York opens AI regulatory registration platform; OpenAI and Anthropic must report incidents within 72 hours

    New York State has officially opened the RAISE Act regulatory registration platform. Frontier AI developers with annual revenue exceeding $500 million must register by Jan 1, 2027, publish their safety frameworks, and report major safety incidents within 72 hours; violations carry fines of up to $3 million.

  • Related companiesAnthropicThird-party report

    Anthropic launches Claude Whiteboards, Sheets, and Data

    Anthropic has begun rolling out Claude Whiteboards, Claude Sheets, and Claude Data for all paid plans, with support on web, desktop, and mobile; Whiteboards and Sheets are in beta, while Claude Data is now generally available.

  • Related companiesBaidu ERNIEThird-party report

    ERNIE's new model builds Backrooms entirely from code, switches 20 art styles in 116 seconds

    Baidu's new ERNIE model can generate Backrooms scenes purely from code, switching between 20 art styles in 116 seconds; reports say the model will soon be open-sourced.

  • Related companiesAnthropicThird-party report

    Anthropic launches IPO roadshow, seeks to raise up to $64.35 billion

    Anthropic has launched its IPO roadshow, planning to issue 825 million shares and raise up to $64.35 billion. Combined with a $26 billion private placement, total fundraising reaches $90.4 billion, with a valuation range of $1 trillion to $1.4 trillion. Google, Nvidia and others are participating in the subscription.

  • Related companiesxAIThird-party report

    Grok accused of generating child sexual abuse imagery, xAI faces multiple class-action lawsuits

    According to Bloomberg, Grok, the chatbot from Elon Musk's xAI, is accused of altering ordinary photos into child sexual abuse material, with a victim participating in a class-action lawsuit against xAI under the name Jane Doe 5.

  • Related companiesOpenAIAnthropicThird-party report

    TSMC approves up to $500 million investment each in Anthropic and OpenAI

    TSMC's board approved two investment plans: subscribing to up to $500 million of Anthropic common stock at its IPO price, and investing up to $500 million to subscribe to OpenAI preferred stock. Anthropic also announced plans for an IPO at $70 to $78 per share.

  • Related companiesGoogle AIThird-party report

    Google responds to AI's impact on its advertising giant business

    Reports say Google Search ads are seeing an unusual phenomenon of unspent budgets, with advertisers reporting insufficient traffic; Google is responding to AI's impact on its advertising business.

  • Related companiesAlibaba QwenThird-party report

    Alibaba releases Qwen Code 1.0, integrating Qoder open source

    Alibaba officially announced the launch of Qwen Code 1.0, integrating and open-sourcing the Harness capabilities refined in Qoder. Released with the Preview version, it provides developers with an Agent runtime foundation that can be embedded into their own products.

  • Related companiesOpenAIThird-party report

    OpenAI acquires AI travel agent company Odessia

    OpenAI has acquired Odessia, an AI travel agent company founded just 1 year ago. Founder Francis Davidson has joined OpenAI, but OpenAI has not yet announced any new products or plans in the travel sector.

1 more item in this issue is still being translated and will appear once it's ready.

Generates a long image you can save and share.