AI Daily: Kolors 3.0 Released; Alibaba's Large Model Brand Officially Renamed Qwen; Mistral AI Releases Voxtral Transcribe 2 Voice Model

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 technical trends and innovative AI product applications.

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

1. World-first "Subject Reference": Clever AI 3.0 officially launched, 15-second long video opens the era of AI directors

The release of Clever AI 3.0 marks a new era in AI video creation. Through integrated multi-modal input and output, it redefines the AI video creation workflow, allowing everyone to become a director.

【AiBase Highlights:】

🎥 World-first subject reference function, accurately identifies the main character, props, or scene features.

🎬 Intelligent storyboarding and ultra-long generation, achieving cinematic storytelling.

🖼️ Supports multiple languages and dialects, achieving perfect synchronization of lip movements, expressions, and voice tones.

2. Alibaba AI Strategy Upgraded Again: The Large Model Brand Officially Renamed "Qwen"

Alibaba Group announced that the company's AI technology and core brand will be unified under the name "Qwen", aiming to eliminate market confusion caused by multiple brands, marking a new stage of high focus in Alibaba's AI branding. At the same time, SpaceX submitted an application for deploying a million-satellite system, drawing industry attention.

【AiBase Highlights:】

🧠 Unifying the Qwen large model brand to "Qwen" improves market recognition consistency.

🌐 Globally using the English name Qwen strengthens the international brand image.

🛰️ SpaceX submitted an application for deploying a million-satellite system, attracting widespread attention.

3. Latency Below 0.2 Seconds! Mistral AI Launches Voxtral Transcribe 2 Speech Model, Supporting Real-Time Chinese Transcription

Mistral AI launched the Voxtral Transcribe 2 speech-to-text model series, optimized for different application scenarios, solving the issues of high latency and cost in voice interaction. Voxtral Realtime achieves extremely low latency, while Voxtral Mini is known for its cost-effectiveness and long audio processing capabilities. Both models support 13 languages including Chinese, with competitive pricing strategies suitable for various voice interaction scenarios.

【AiBase Highlights:】

⚡ Extremely low latency: The Voxtral Realtime model reduces transcription latency to within 200ms, supporting real-time audio transcription, and the model weights are open-source.

🏆 Cost-effective: The Voxtral Mini version outperforms similar products like GPT-4o mini in accuracy, supports 3-hour long recordings, and offers significant price advantages.

🌐 Multilingual support: All models in the series support 13 languages including Chinese natively, widely adapting to globalized voice office and real-time interaction scenarios.

4. Shanghai AI Lab Releases the Scientific Multimodal Model ShuSheng Intern-S1-Pro

The Shanghai AI Lab released the globally largest open-source parameter scientific multimodal model "ShuSheng Intern-S1-Pro". Based on the SAGE technology architecture, the model has 1 trillion parameters and demonstrates internationally leading scientific capabilities, especially in logical reasoning and research intelligent agents.

【AiBase Highlights:】

✨ ShuSheng Intern-S1-Pro is the world's largest open-source parameter scientific multimodal model, with 1 trillion parameters.

🧠 The model performs well in complex interdisciplinary evaluations, with logical reasoning abilities equivalent to gold medalists in Olympic competitions.

🚀 It uses a mixture-of-experts architecture (MoE) to optimize computational efficiency and reduce resource consumption, laying the foundation for training ultra-large-scale models.

5. Gemini User Base Surpasses 750 Million: Google's Financial Report is Impressive, Keeping up with ChatGPT

Alphabet, the parent company of Google, released its fourth-quarter 2025 financial report, where the AI business performed impressively. The monthly active users of Gemini have exceeded 750 million, showing remarkable growth momentum. At the same time, Google introduced a low-cost subscription plan called Google AI Plus and launched its own chip Ironwood, driving the company's annual revenue to break through the $40 billion mark for the first time, marking a complete transformation into an "AI-first" enterprise.

【AiBase Highlights:】

📈 Gemini user base has surged to 750 million, showing strong growth momentum.

💰 Google introduced a low-cost subscription plan Google AI Plus to attract price-sensitive users.

🚀 Self-developed chip Ironwood and Gemini 3 helped Alphabet's annual revenue exceed $40 billion for the first time.

6. Support AI to Eliminate Screen Moiré Pattern! Huawei Mate 80 Series Officially Released HarmonyOS New Firmware

Huawei Mate 80 series phones were updated to HarmonyOS 6.0.0.130SP17, introducing the AI screen moiré pattern elimination function, improving image processing capabilities, and optimizing Starlight and positioning performance.

【AiBase Highlights:】

✨ Introduced AI photo editing - screen pattern elimination function, accurately removes moiré patterns, enhancing image clarity.

📶 Optimizes Starlight audio connection stability, enhancing Bluetooth experience.

🧭 Improves navigation positioning accuracy, optimizes mobile network connection performance.

7. First in Olympic History! Milan Winter Olympics Announces the Use of Alibaba's Official "Qwen" Large Model

The article introduces that the 2026 Milan Winter Olympics will use Alibaba's official Qwen large model, pushing the Olympic technology toward the intelligent era, and mentions Bilibili's collaboration with the Spring Festival Gala, showcasing the extensive application of AI in top IP.

【AiBase Highlights:】

🧠 The first official large model in Olympic history has been introduced, marking the beginning of the intelligent Olympics.

📺 Bilibili collaborates with the Spring Festival Gala to enhance the interactive experience of bullet comments.

🚀 AI is deeply reshaping the business logic and user experience of global top IPs.

8. China's Generative AI User Base Exceeds 600 Million: Penetration Rate Over 40%, Computing Power Level Ranks Among Global Leaders

China's generative AI user base has grown rapidly, with a significant increase in penetration rate. At the same time, computing power infrastructure has made breakthrough progress, providing a solid foundation for the widespread application of AI technology.

【AiBase Highlights:】

📈 China's generative AI user base has reached 602 million, growing by 141.7% from the end of 2024, with a penetration rate of 42.8%.

💻 A total of 42 ten-thousand-card smart computing clusters have been built nationwide, with the total scale of intelligent computing power exceeding 1590 EFLOPS, ranking among the global leaders.

🚀 Generative AI is accelerating its integration into daily life and production, becoming an important engine for digital transformation and high-quality economic development.