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 innovative AI product applications.
Fresh AI products Click to learn more: https://app.aibase.com/zh
1. Douyin E-commerce Integrates Doudao into the Store Settlement Sequence, GEO Enters the Era of Transaction Attribution
Douyin E-commerce has integrated the large model application "Doudao" into the store settlement sequence, marking the commercial calculation of AI-driven orders and building a comprehensive cost model of "category + channel", providing an industry example for the commercialization of large models.
AiBase Summary:
🛒 Douyin E-commerce integrates the large model application "Doudao" into the technical service fee rules, marking the commercial calculation of AI-driven orders.
📊 The platform has adjusted the base technical service fee rates for specific categories and built a comprehensive cost model of "category + channel".
🔍 Generative AI search optimization (GEO) has shifted from exposure logic to transaction attribution and cost accounting stages.
2. Anthropic Re-extends: Fable5 Access Rights Extended to July 19, Retaining Users and Recovering Reputation
Anthropic announced the extension of access rights for its flagship model Claude Fable5 to July 19, while maintaining the rate cap for Claude Code. This move aims to address competition in the AI industry and restore brand image, while allowing users to enjoy discounts during the extension period. [AiBase Summary:]
📅 Fable5 access rights extended to July 19 to retain developer users.
🛡️ Anthropic faces security and compliance issues, and brand image recovery still needs time.
🔄 Usage of Fable5 counts as half of the subscription limit, without additional fees.
3. SenseTime Opensources SenseNova-Vision Unified Vision Large Model, Outperforming in Four Core Visual Tasks
SenseTime released and open-sourced the SenseNova-Vision unified vision large model, which performs exceptionally well in multiple visual tasks, showing capabilities surpassing existing models. It also open-sourced a large dataset, providing important support for the industry.
AiBase Summary:
🧠 The single model leads comprehensively in four core visual fields, outperforming specialized models.
📊 Achieves comprehensive superiority over Vision Banana in detection, segmentation, depth, and other tasks.
📦 Open-source model and a visual instruction corpus containing 50 million samples, promoting technological popularization.
4. Claude Code Evolves Again! Built-in Browser Allows AI to "Browse Websites" Autonomously
Anthropic brought significant upgrades to Claude Code, adding a built-in browser window feature that allows developers to directly access and operate external websites within the application, improving work efficiency. At the same time, multiple security review mechanisms ensure the safety of AI behavior, and enterprises can manage AI's access range through white lists to meet different security needs in various scenarios.
AiBase Summary:
🧠 Built-in browser window enhances developer efficiency.
🔒 Security review mechanism ensures user control.
🌐 White list management meets enterprise-level security needs.
5. Nubia AI Pet Robot iMoochi Scheduled for Release on July 17, Focuses on Emotional Support and Lifelike Interaction
The Nubia AI pet robot iMoochi focuses on emotional support, offering a unique companionship experience for urban singles through lifelike interaction design and personalized growth features.
AiBase Summary:
🧸 iMoochi provides a warm companionship experience through lifelike interaction design.
🧠 Powerful hardware configuration supports personalized growth and voiceprint identification functions.
🎨 Five personality choices, meeting different users' emotional needs.
6. Surging Demand for Premium Models, OpenAI Temporarily Lifts GPT-5.6Sol Compute Limitations
To cope with the surge in demand for GPT-5.6Sol, OpenAI temporarily removed the 5-hour usage limit for Plus, Pro, and Business subscribers, and optimized model performance to reduce compute and token consumption. This measure not only addresses short-term traffic pressure but also reflects the trend of the large model industry moving towards engineering and cost reduction.
AiBase Summary:
🧠 OpenAI temporarily lifts the usage restrictions for GPT-5.6Sol to address the surge in market demand.
🚀 Optimizes model performance to reduce compute and token consumption, enhancing task capacity.
📊 The industry trend shifts from "parameter arms race" to "practical engineering and cost reduction."
7. Ant Group Releases Two Open-Source Security Models, Putting a "Sutra" on Large Models and Intelligent Agents
Ant Group announced the open-sourcing of two intelligent agent security models, SingGuard-NSFA and multi-modal security barrier SingGuard, aiming to provide more comprehensive underlying security capabilities for large models and intelligent agents. These models offer comprehensive solutions for challenges such as intelligent agent behavior control, permission management, and multi-modal content security.
AiBase Summary:
🧠 The intelligent agent security barrier SingGuard-NSFA provides real-time security detection for intelligent agent operations, covering 7 major categories, 28 medium categories, and 185 specific scenarios.
🛡️ The multi-modal security barrier SingGuard can identify complex attacks hidden in text, images, and other modalities, and supports dynamic loading of natural language security rules at runtime.
🌐 Ant Group continues to participate in AI security standards and governance system construction, promoting AI security capabilities from technological innovation to industrial practice.
8. StepZen Launches Step Edge Series Terminal Models, Achieving Efficient Multimodal Processing Locally
StepZen launched the Step Edge series terminal models, achieving efficient multimodal processing locally. The model can process text, images, and audio locally, supporting screen understanding and speech recognition, while featuring low latency and high privacy security.
AiBase Summary:
🧠 The Step Edge series models can process text, images, and audio locally, supporting screen understanding and speech recognition functions.
📶 The series models have a call latency as low as 0.1 seconds, significantly reducing reliance on network and cloud computing for devices.
🚀 The official statement says that the Step Edge series models rank first in 29 core evaluation indicators, demonstrating leading edge-side multimodal capabilities.
