Welcome to the AI Daily section! This is your daily guide to exploring the world of artificial intelligence. Every day, we bring you the hottest topics in the AI field, focusing on developers, helping you understand technological trends and innovative AI product applications.
Discover fresh AI products by clicking here: https://top.aibase.com/
1. The most powerful model, Llama3, officially released, reaching GPT-4 level
Llama3 is Meta's latest open-source model, available in 8 billion and 700 billion parameter sizes, expected to be officially released in July. This model boasts multimodal capabilities, integrates new computer coding features, and will be applied to Meta's smart glasses and other devices. Compared to its predecessor, Llama2, it has made significant strides in reasoning, code generation, and instruction tracking. Meta has partnered with Google to integrate real-time search results into the AI assistant, enhancing user experience.
AiBase Highlights:
✨ Llama3 includes 8 billion and 700 billion parameter sizes, with multimodal capabilities, to be applied in devices like Meta's smart glasses.
🚀 Llama3 has made significant progress in reasoning, code generation, and instruction tracking, with breakthrough performance.
💡 Meta and Google collaborate to integrate real-time search results into the AI assistant, enhancing user experience.
Experience address: https://www.meta.ai/
Model download: https://top.aibase.com/tool/meta-llama-3
Hugging Chat: https://huggingface.co/chat/conversation/6621c6d78f6a88b6b6084ac4
2. Midjourney introduces new feature Room, allowing users to create images together in a chatroom
Midjourney has recently launched an interesting new feature—Room, providing a collaborative and social platform where users can create and share images together and participate in real-time chats. Room promotes interaction and cooperation among users, inspires inspiration and creative exchanges, and enriches the Midjourney experience.
AiBase Highlights:
🎨 Users can create and share images together, exchanging ideas and feedback in real-time chats.
💬 Room offers voice and text chat features, facilitating discussions on the creative process and sharing ideas.
🤝 Images generated on the left can be easily referenced in the chat, promoting interaction and cooperation among users.
Details link: https://www.midjourney.com/showcase
3. Baidu Library launches "Smart Comic" and "Smart Sketchbook" features
Baidu Library announced the new "Smart Comic" and "Smart Sketchbook" features at the 2024 Create2024 Baidu AI Developer Conference, offering users a creation experience of generating comics or sketchbooks with a single sentence, while also promoting the development of cross-modal content creation. Users can earn revenue through their creations, enhance parent-child communication, and participate in comic creation competitions and community exchanges. Baidu Library has partnered with Xuanji Technology to strive to create a globally leading smart comic solution.
AiBase Highlights:
✨ Users can create comics or sketchbooks with just one sentence, achieving a full-chain creation path.
🎨 Baidu Library's "Smart Comic" feature optimizes story generation, offering multiple styles and character choices, creating high-quality works.
📚 The "Smart Sketchbook" feature allows parents to create exclusive stories for their children, enhancing parent-child communication, providing picture-text-voice integrated video experiences.
Details link: https://wenku.baidu.com/pcactivity/makeAppointment?_wkts_=1713506418546&
4. AI video editing tool Captions supports lip-syncing and AI-auto added sound effects
This AI video editing tool called Captions revolutionizes the video editing process by providing a series of automated features that enhance the visual and auditory effects of videos. Users can choose subtitle templates, add sound effects, stickers, and other elements to make videos more attractive. Captions also supports AI eye contact, AI noise reduction, AI lip synchronization, and AI color grading, making videos more professional and exquisite. Its web version is convenient for users to use, and plans to add automatic subtitle translation in the future to attract more global users.
AiBase Highlights:
✨ Revolutionary video editing tool, automatically identifies important clips to create short videos, simplifying the editing process.
🎬 Provides automated features to enhance visual and auditory effects, users can choose subtitle templates, add sound effects, stickers, and other elements.
💡 Supports AI eye contact, noise reduction, lip synchronization, color grading, and other functions to enhance the professionalism and attractiveness of videos.
Details link: https://top.aibase.com/tool/captions
5. You can now use open-source models on your phone! Hugging Face releases iOS app Hugging Chat
Hugging Face has recently launched an iOS client app called "Hugging Chat", bringing a convenient user experience. Users can now access and use multiple open-source models hosted on the Hugging Face platform on their phones. The app responds quickly, providing users with a smooth experience, although Chinese support needs to be improved. This innovative move once again demonstrates Hugging Face's leading position in the field of language models and its attention to user needs.
AiBase Highlights:
📱 Users can now access and use multiple open-source models hosted on the Hugging Face platform on their phones.
⚙️ The app provides six models, including Command R+, Zephyr-14b, Mixtral8x7B, etc., although Chinese support is not good, it is still popular.
🚀 The application responds very quickly, providing users with a smooth experience.
Details link: https://apps.apple.com/us/app/huggingchat/id6476778843
6. Nothing announces that its headphones will support ChatGPT
Nothing announces deeper integration of ChatGPT into its smartphones and headphones, providing faster services to users. Users can interact with AI tools quickly through voice without manual operation. After the update, users can use ChatGPT on Nothing headphones, enhancing the experience of smart assistants. The company plans to go beyond this and embed system-level access points to ChatGPT in Nothing OS, improving the user experience of smartphones.
AiBase Highlights:
🔊 Users can interact with AI tools quickly through voice without manual operation.
📱 After the update, users can use ChatGPT on Nothing headphones, enhancing the experience of smart assistants.
🚀 Nothing plans to integrate ChatGPT more deeply into its smartphones and headphones, providing faster services to users.
7. Stability AI announces layoffs of 20 people, about 10% of its workforce
Stability AI announces layoffs of 20 people, the decision to cut costs, strengthen support for investors and partners, and promote the team to continue developing and releasing innovative products. The company has recently encountered personnel turnover issues, the founding CEO has resigned, and is facing alleged infringement litigation. The layoff decision was made after a turbulent few weeks following the CEO's departure.
AiBase Highlights:
📉 The company lays off 20 people, accounting for about 10% of the total number of employees, aiming to cut costs, strengthen support, and promote product development.
💼 The company has recently encountered personnel turnover issues, the founding CEO has resigned, and is facing alleged infringement litigation.
⚖️ The company's layoff decision is part of a strategic plan, and the layoff department has not yet been clearly identified.
8. OpenAI's image generator DALL-E2 ceases service, DALL-E3 takes its place
OpenAI announces the shutdown of the DALL-E2 image generator service, replaced by DALL-E3, which integrates with ChatGPT, offering a simpler interface and better quality images. Users should note that DALL-E2 will no longer accept new customers and recommend using DALL-E3 for image generation.
AiBase Highlights:
🔍 DALL-E3 integrates with ChatGPT, offering a simpler interface and better quality images.
💡 DALL-E3 provides higher quality images, improved prompt compliance, suitable for ChatGPT Plus, Team, and Enterprise as well as OpenAI API.
🛑 DALL-E2 will no longer accept new customers, users cannot create new images in the DALL-E2 interface, it is recommended to download particularly liked content for safety.
9. The Cyberspace Administration of China confirms that 116 AI large models have been filed
The Cyberspace Administration of China released the "Announcement on the Filed Information of Generative AI Services", confirming that 116 AI large models have passed the filing procedures, meaning they comply with relevant specifications and can be legally applied within China. This initiative promotes the innovative development and standardized application of generative AI services, which is of great significance to the development of the AI field.
AiBase Highlights:
📝 116 AI large models have been confirmed through filing, legally applicable within China.
📝 Filed AI large models cover multiple fields and application scenarios.
📝 Generative AI applications that have gone live need to publicize filing information, increasing information transparency.
10. Who is the best among domestic large models? Tsinghua's latest evaluation report of 14 LLMs is out, with GLM-4 and Wenxin 4.0 in the first echelon
This article summarizes Tsinghua University's Basic Model Research Center's latest "SuperBench Large Model Comprehensive Capability Evaluation Report", which comprehensively evaluates 14 LLMs at home and abroad. In the evaluation, domestic models such as GLM-4 and Wenxin 4.0 have performed brilliantly, gradually narrowing the gap with international top models.
AiBase Highlights:
