Welcome to the 【AI Daily】 column! Here is your daily guide to exploring the world of artificial intelligence. Every day, we present you with the hottest topics in the AI field, focusing on developers, helping you gain insights into technology trends and understand innovative AI product applications.

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

1. Baidu's Ernie Bot will be completely free starting April 1, with deep search functionality launched

The official announcement from the Ernie large model introduces two significant updates that will greatly enhance user experience. Starting April 1, Ernie Bot will be completely free, allowing users to access the latest model without barriers on both PC and app platforms. Additionally, the deep search feature will also be available for free, supporting multi-modal input and output, providing expert-level content responses, and handling multi-scenario tasks.


image.png

【AiBase Summary:】

🆓 Ernie Bot will be fully free from April 1, allowing users to experience the latest model without barriers on PC and app platforms.

🔍 The newly launched deep search feature will also be available for free on the same day, equipped with powerful thinking and planning capabilities.

🌐 Deep search supports multi-modal input and output, enhancing the user search experience and meeting diverse scenario needs.

2. OpenAI is set to make a big move: GPT-5 will be available for free and unlimited use, expected to launch in a few months

OpenAI CEO Sam Altman recently revealed significant updates on GPT-4.5 and GPT-5, marking another major advancement for the company in the field of artificial intelligence. GPT-5 is expected to launch in a few months, with free users enjoying unlimited conversations, while Plus and Pro users will have access to smarter features. GPT-4.5 is referred to as the Orion model, indicating OpenAI's future direction.


image.png

【AiBase Summary:】

🆓 Free users will be able to have unlimited conversations with GPT-5, experiencing the latest AI technology.

✨ Plus and Pro users will enjoy smarter features, including voice, canvas, and deep research capabilities.

🔍 GPT-4.5, known as the Orion model, will be the last non-chain-of-thought model, guiding OpenAI's future development.

3. Kuaishou launches CineMaster: a video version of ControlNet, achieving precise 3D perception text-to-video generation

Kuaishou has launched CineMaster, an innovative text-to-video generation framework with 3D perception capabilities, similar to a video version of ControlNet. This framework allows users to precisely control the position of objects and camera movements in the video using various control signals, greatly enhancing creative freedom. Users can generate videos not only through text prompts but also by utilizing depth maps, camera trajectories, and object labels for detailed adjustments, creating more personalized and creative works.


image.png

【AiBase Summary:】

🎬 CineMaster is a brand new text-to-video generation framework with 3D perception capabilities, providing unprecedented creative freedom.

📏 Users can precisely control the position of objects and camera movements in the video using control signals like depth maps, camera trajectories, and object labels.

🔍 Kuaishou also provides a process for extracting 3D bounding boxes and camera trajectories from large-scale videos, offering strong data support for CineMaster's training and application.

4. Tencent Yuanbao supports Deepseek R1 for online searching of public account content

Tencent Yuanbao recently underwent a significant upgrade, officially supporting the DeepSeek R1 model and introducing the ability to search public account content online. This update significantly improves users' information retrieval efficiency, and with voice and text input, Yuanbao can provide the latest and most authoritative information, enhancing the accuracy and timeliness of responses.


image.png

【AiBase Summary:】

🧠 Yuanbao supports the DeepSeek R1 model, enhancing deep thinking and logical reasoning capabilities.

📱 The new online search feature supports voice and text input, integrating various information sources.

✍️ Provides AI reading and writing features to help users quickly obtain document highlights and creative inspiration.

5. Adobe Firefly AI video generation tool public beta: easily convert text and images into short videos

Adobe recently launched the public beta version of Firefly Video, allowing users to convert text and images into short videos of up to 5 seconds in 1080p. The tool features a user-friendly interface and supports various customization options, greatly enhancing the flexibility of video creation. In addition, Firefly Video also includes AI translation capabilities, supporting over 20 languages, providing convenience for multilingual content creators.


image.png

【AiBase Summary:】

🎥 The Adobe Firefly Video generator can convert text and images into short videos of up to 5 seconds in 1080p, allowing users to make multiple custom settings.

💡 The newly launched Firefly Standard and Firefly Pro subscription plans offer different video generation quotas and translation features.

🌍 The AI translation function supports over 20 languages, providing convenience for multilingual content creators, ensuring translated videos match the original audio.

Details link: https://www.adobe.com/de/products/firefly.html

6. Alibaba releases Animate Anyone 2: easily achieve character replacement in videos and seamless transfer of actions and expressions

Alibaba recently launched Animate Anyone 2, a revolutionary technology that allows precise character replacement and seamless transfer of actions in videos. Users simply provide the original video and target character images, and the system can replace character appearances in real-time while preserving motion trajectories. This technology excels in character consistency and detail retention, enhanced by dynamic bone binding and object guiders for improved interaction between characters and their environment.


image.png

【AiBase Summary:】

🌍 Animate Anyone 2 achieves realistic interactions between characters and environments by capturing environmental features.

🎨 The new technology introduces object guiders and spatial blending to enhance the interaction effects between characters and objects.

🚀 Compared to similar technologies, Animate Anyone 2 shows significant advantages in detail and consistency.

Details link: https://humanaigc.github.io/animate-anyone-2/

7. The new video generation model Magic 1-For-1: generates a one-minute video in one minute

The Magic 1-For-1 model developed by the DA-Group-PKU team is known for its efficient image-to-video generation technology, capable of generating a one-minute video clip in just one minute. This technology significantly enhances video generation efficiency by optimizing memory usage and reducing inference delays. The model breaks down the video generation process into two key sub-tasks: text-to-image generation and image-to-video generation, improving training efficiency and generation accuracy.


image.png

【AiBase Summary:】

📹 Efficient generation: The Magic 1-For-1 model can generate a one-minute video in one minute, optimizing memory usage and reducing inference delays.

📥 Open resources: The team has released a technical report, model weights, and code, welcoming developers and researchers to contribute.

💻 Flexible inference: Supports single GPU and multi-GPU inference settings, allowing users to choose the suitable operation mode based on their needs.

Details link: https://github.com/DA-Group-PKU/Magic-1-For-1

8. Lei Jun discusses DeepSeek: highly admires them and encourages using AI to improve work efficiency

During the Lantern Festival, Xiaomi CEO Lei Jun interacted with netizens through a live stream, sharing information about the upcoming Xiaomi SU7 Ultra and Xiaomi 15 Ultra, and discussing the impact of AI technology. He emphasized the importance of AI in daily life and encouraged everyone to learn how to use AI tools to improve work efficiency. Lei Jun expressed admiration for the DeepSeek team, believing that the widespread application of AI technology will provide individuals with a competitive advantage.


image.png

【AiBase Summary:】

🍽️ Lei Jun live streamed in the Xiaomi cafeteria, interacting with netizens and sharing new product information.

🤖 Emphasized that AI technology has profoundly influenced daily life, urging everyone to learn how to use AI tools.

💡 Mentioned that the widespread application of AI will provide individuals with a competitive advantage, and we need to quickly adapt to the new era.

9. Accelerating AI layout! Meta seeks to acquire AI chip company FuriosaAI

Meta is in talks to acquire the South Korean chip startup FuriosaAI to enhance its artificial intelligence hardware infrastructure. This move aims to address the growing demand for AI, as FuriosaAI focuses on developing chips that accelerate the operation of AI models, suitable for Meta's text generation models like Llama2 and Llama3. Meta plans to invest up to $65 billion in AI initiatives in 2023, demonstrating its commitment to the future development of AI technology.


image.png

【AiBase Summary:】

🌟 Meta is in talks to acquire FuriosaAI to enhance its AI hardware infrastructure.

💡 The chips developed by FuriosaAI can accelerate the operation of Meta's text generation models.

💰 Meta plans to invest up to $65 billion in AI initiatives in 2023.

10. Anthropic expects revenue to reach $34.5 billion by 2027

Anthropic has demonstrated strong growth potential in its latest financial forecast, expecting revenue to reach $3.7 billion by 2025, with cash outflows reduced by nearly half. With the surge in demand for artificial intelligence, the company is facing significant market opportunities. Management is optimistic about future development, predicting revenue will soar to $34.5 billion by 2027, relying on continued R&D investment and flexible market responses. Industry analysts generally view Anthropic's position in the competitive market as increasingly important.


image.png

【AiBase Summary:】

🌟 By 2025, Anthropic expects revenue to reach $3.7 billion, with cash outflows reduced by nearly half.

📈 The company anticipates revenue to soar to $34.5 billion by 2027, demonstrating strong growth potential.

🤝 Anthropic will continue to increase R&D investment and actively explore collaborations with other companies to enhance market competitiveness.