Welcome to the "AI Daily" section! This is your guide to exploring the world of artificial intelligence every day. Every day, we present you with the latest content in the AI field, focusing on developers to help you understand technology trends and learn about innovative AI product applications.

Fresh AI products Click to learn more:https://app.aibase.com/zh

1. Google Cloud Launches General-Purpose Gemini Agent, Capable of Self-Disassembling and Executing Complex Multi-Step Tasks

Google Cloud officially launched a new Gemini agent at the "Gemini at Work 2026" conference. As a general-purpose work agent, this product will completely change the way enterprises collaborate and improve productivity through a one-stop interface. It can not only handle tasks such as answering questions, knowledge consolidation, and content creation, but also deeply call internal enterprise tools and data to achieve full-process automation and deliver the final product to Gmail, Docs, or development tools. In addition, the Gemini agent also has ecological compatibility and industry-specific applications, while providing comprehensive enterprise-level security control features.

image.png

AiBase Summary:

🧠 The Gemini agent can self-disassemble and execute complex multi-step tasks

🌐 Full upgrade of ecological compatibility, supporting multiple models and third-party services

🔒 Comprehensive enterprise-level security control functions to meet compliance and risk control needs

2. OpenAI Launches GPT-6.1 Sol Ultrafast Version, Speed Up to 8 Times Faster

OpenAI officially released the GPT-6.1 Sol Ultrafast version, significantly improving performance and optimizing pricing strategies, while supporting global data compliance requirements, providing more efficient service for different users.

image.png

AiBase Summary:

🚀 The GPT-6.1 Sol Ultrafast version is now fully available across API, Codex, and ChatGPT Work platforms, with speed up to 8 times faster.

💰 Price is 6 times that of the standard version, with input cost of $12 per million tokens and output cost of $60 per million tokens.

🌍 Supports data residency compliance services in the US and EU, suitable for Pro500 users, enterprise versions, and educational versions.

3. Doubao App Launches Life Payment Function, Accelerating Layout of High-Frequency Daily Life Scenarios

Doubao App launches life payment function, accelerating the layout of high-frequency daily life scenarios, marking the transformation of large model products from text generation to real scenario operations.

image.png

AiBase Summary:

💡 Doubao App adds the "Life Payment" function, allowing users to pay water, electricity, and gas bills directly via voice or typing.

🚀 Doubao is gradually building multi-dimensional life scenario capabilities including travel services and life payments.

🌐 Doubao connects with basic public services through natural language interaction, expanding the practical boundaries of generative AI.

4. Zhipu AI Reveals New Large Model Roadmap: Parameter Count Breaks Trillion, Dual-Track Technology Path Aims for AGI

Zhipu AI's new large models GLM-5.4 and GLM-5.5 achieved significant breakthroughs in parameter scale, entering the trillion level, and through the dual-track technology path, significantly improved the efficiency and effectiveness of the model, promoting domestic large models towards General Artificial Intelligence (AGI).

image.png

AiBase Summary:

🧠 Parameter count exceeds the trillion level, enhancing complex reasoning and multimodal understanding capabilities

🚀 Adopting a dual-track technical path to optimize training and inference efficiency

🌐 Promoting domestic large models toward General Artificial Intelligence (AGI)

5. Kuaishou Releases "AI Mod" Video Governance Announcement: Disposed 2,544 Pieces of Violating Content in September, Focused on Restricting Misuse of Classic Works

Kuaishou Security Center issued a special governance announcement regarding "AI Mod" videos, stating that the platform disposed of 2,544 pieces of violating content in September. This governance focuses on improper marriage-related associations or cross-time-space interactions between classic film characters such as "Journey to the West" and "Dream of the Red Chamber" and popular game characters, indicating that the platform's compliance review of generative AI derivative content has been significantly intensified. The National Radio and Television Administration has launched a special governance campaign for "AI Mod" videos, clearly pointing out that cultural appropriation and alteration based on the Four Great Classical Novels may lead to cultural identity confusion, impacting cultural recognition. At the same time, AI derivative content also faces multiple legal compliance risks, including copyright, right of portrait, and protection of the integrity of works.

image.png

AiBase Summary:

🎮 AI-generated content leads to inappropriate modifications of classic film characters

📚 The National Radio and Television Administration has launched a special governance campaign for "AI Mod" videos

⚖️ Generative AI derivative content faces multiple legal compliance risks

6. OpenAI Revises Annual Revenue Forecast to $50 Billion, IPO Plan Delayed to 2027

The article points out that OpenAI disclosed its annual revenue of $50 billion to investors, lower than the previously expected $70 billion. This data adjustment has sparked discussions about the commercialization and expenditure sustainability of leading large model companies. Despite raising $12.2 billion in March, OpenAI's 2025 financial report shows revenue of about $13 billion, with R&D and operational expenses far exceeding income. Therefore, OpenAI's IPO plan has been delayed to early 2027.

image.png

AiBase Summary:

💰 OpenAI's annual revenue was adjusted down from $70 billion to $50 billion, drawing market attention.

📊 Differences in accounting standards have led to discrepancies in revenue data, with Anthropic including cloud partner sales while OpenAI does not.

⏳ OpenAI has postponed its IPO plan to 2027, reflecting commercial pressure and revised revenue growth expectations.

7. Apple Announces "Welcome Home" Event on October 13, Fully Focuses on Smart Home AI Hub

Apple announced that it will hold a smart home product launch event titled "Welcome Home" on October 13 in New York. This is the first major strategic move in the smart home field by new CEO John Ternus, marking Apple's full revitalization of its market strategy in this area. The core highlights of this event include a new form of smart home hub hardware, as well as an upgraded version of the Apple TV 4K and an updated HomePod mini smart speaker. All new products are deeply integrated with the latest Siri AI technology, aiming to redefine the user experience of the entire home smart ecosystem.

image.png

AiBase Summary:

🍎 Apple will launch a new smart home hub hardware with a 6-inch square display design.

📺 The new Apple TV 4K and upgraded HomePod mini smart speaker will also be unveiled simultaneously.

🤖 All new products are deeply integrated with the latest Siri AI technology, redefining the user experience of the whole-home smart ecosystem.

8. An Infinite Canvas Holds All Creations: Doubao Work Updates, Images, Videos, HTML Automatically Merged into One Canvas

Doubao Work launched dual-line updates, adding a canvas feature and integrating the Seedream5.0Flash image model, while also launching the Doubao 2.1Lite model to enhance office efficiency and multimodal generation capabilities.

image.png

AiBase Summary:

🎨 The canvas feature supports complex creative tasks, allowing users to place materials, plans, and creative results on the same infinite canvas.

🤖 The Doubao 2.1Lite model focuses on saving quotas and faster delivery, suitable for frequent office scenarios.

🖼️ The integration of the Seedream5.0Flash image model provides more style and visual expression options for image generation.