Welcome to the AI Daily section! This is your daily guide to exploring the world of artificial intelligence. Each day, we bring you the hottest topics in the AI field, focusing on developers to help you understand technological trends and discover innovative AI product applications.
Fresh AI Products Click to Learn More: https://top.aibase.com/
1. OpenAI Releases GPT-4o mini, API Price Cut by 60%!
OpenAI has launched GPT-4o mini, dubbed the "economically suitable male" in the AI field, with a 60% price reduction and superior performance. This model excels in both cost and performance, paving the way for the AI revolution.
AiBase Highlights:
💡 GPT-4o mini's price cut by 60%, with 15 cents per million input tokens and 60 cents per output tokens, offering excellent cost-effectiveness.
💡 Supports text and visual APIs, with future expansions to image, video, and audio input-output, offering powerful and diverse functionalities.
💡 Outstanding performance in academic tests, surpassing GPT-3.5 Turbo, supporting multimodal reasoning with a 128K token context window.
2. iFlytek Launches Spark Pro-128K Large Model
iFlytek's Spark Pro-128K large model boasts exceptional long-text processing capabilities, significantly enhancing the value of AI large models in practical applications. The model not only handles long texts but also features content tracing, improving the reliability of responses. Developers can claim 5 million tokens for free to experience and utilize this large model's capabilities.
AiBase Highlights:
🔥 Long-text processing capabilities enhance the value of AI large models
💡 Content tracing feature improves response reliability
🚀 Developers can claim 5 million tokens for free to experience the large model's capabilities
3. DeepL Launches New Generation Translation AI, Outperforming GPT-4
DeepL's latest generation language model surpasses Google Translate and GPT-4 in translation accuracy, offering users a superior translation experience. Users can easily switch to the new model through the DeepL Pro plan, starting at 1150 yen/month.
AiBase Highlights:
📈 The new generation language model surpasses Google Translate and GPT-4 in translation precision, enhancing translation quality
🔍 The new model performs better in Japanese-English, Simplified Chinese-English, and German-English translations, significantly reducing editing workload
💵 The DeepL Pro plan now supports the new model, allowing users to switch with a button, starting at 1150 yen/month
4. Bilibili's Long Video Summary Tool! Doubao Browser Plugin Launches Podcast and Video Summary Features
In this era of information explosion, the Doubao browser plugin brings a new solution, allowing us to efficiently absorb and organize vast amounts of knowledge. It offers video content summaries, table of contents generation, mind mapping, note-taking functions, and more, greatly enhancing information processing efficiency. Additionally, the plugin supports podcast summaries, article summaries, and translation functions, providing multiple learning tools. Its user-friendly design and versatility make the Doubao browser plugin an essential assistant in our daily learning and work.
AiBase Highlights:
🔍 The Doubao browser plugin provides video content summaries, table of contents generation, mind mapping, and note-taking functions, helping users quickly understand and digest long video content.
🎧 The Doubao browser plugin supports podcast summary functions, allowing users to listen to podcasts and convert them into text summaries, improving information processing efficiency.
📚 The Doubao browser plugin also has article summary and translation functions, with comfortable interaction design suitable for different types of users.
Details Link: Doubao Browser Plugin
5. Groq's Open-Source Llama AI Model Tops the Rankings, Outperforming GPT-4o and Claude in Function Calls
Groq's open-source AI model excels in specific tasks, challenging the proprietary models of tech giants. By training with synthetic data, it addresses common issues like data privacy and overfitting, potentially altering the development path of the AI field and promoting wider AI accessibility and innovation ecosystem cultivation.
AiBase Highlights:
⭐ Groq's open-source AI model outperforms proprietary models of tech giants in specific tasks.
⭐ By using synthetic data for training, Groq challenges common data privacy and overfitting issues in AI model development.
⭐ The release of the open-source model may change the development path of the AI field, promoting wider AI accessibility and innovation ecosystem cultivation.
6. The World's First AI-Translated and Dubbed Film Set to Release, Voice Actors Face Job Crisis!
The world's first AI-translated and dubbed film, Watch The Skies, is set to premiere in American theaters. Flawless's TrueSync technology achieves AI translation and dubbing, offering audiences a more authentic viewing experience.
AiBase Highlights:
🎬 The film Watch The Skies uses Flawless's TrueSync technology for AI-complete translation and dubbing, giving the audience the feeling of watching an English-shot film.
🤖 Flawless's AI dubbing technology perfectly syncs the lip movements of all characters in the film with English, eliminating discomfort for viewers watching dubbed versions.
🌐 Flawless hopes that the TrueSync technology will allow audiences to watch popular shows on streaming platforms without being bothered by mismatched lip movements and brutal scene cuts.
Details Link: Flawless AI Lip Dubbing
7. Video Translation Tool EasyVideoTrans: Fast Conversion with Flexible Editing
In the era of globalization, cross-language communication is crucial. EasyVideoTrans utilizes GPU acceleration technology for rapid video conversion, saving translation time. The operation is simple, customizable, and supports multi-platform use.
AiBase Highlights:
Efficient and Fast Conversion: Utilizes GPU acceleration technology for rapid video conversion.
Personalized Customization: Supports subtitle editing and translation engine selection, allowing users to adjust parameters flexibly.
User-Friendly Operation: Intuitive web interface design, easy to get started, no complex operations required.
Details Link: EasyVideoTrans
8. Google Becomes "Official AI Sponsor of the USA"
Google has become the official AI sponsor of the US team, showcasing its Gemini AI and other AI capabilities during the 2024 Paris Olympics broadcast. Partner NBCUniversal will highlight Google Maps' 3D views and event venue details. Commentators will use Google Search AI to answer Olympic questions, and actors and athletes will also use AI tools like Gemini to participate in various activities. Comedian Leslie Jones will collaborate with Gemini to learn new sports and explore Paris together.
AiBase Highlights:
🏅 Google's Gemini AI will shine during the 2024 Paris Olympics broadcast
🌍 The broadcast will feature Google Maps' 3D views and event venue details
🔍 Commentators will use Google Search AI to answer Olympic questions, and actors and athletes will also use AI tools like Gemini to participate in various activities.
9. New AI "Jailbreak" Tactic! "Past Tense" Prompts Instantly Break Through GPT-4o and Other Six Models
After multiple iterations, large language models (LLMs) excel in natural language processing but also bring risks. Researchers found that simply converting harmful requests into past tense can bypass rejection training, leading to advanced LLMs "jailbreaking." The study reveals the limitations of AI alignment technology and sparks discussions on AI generalization capabilities.
AiBase Highlights:
🔍 Large language models excel in natural language processing but carry risks, such as generating toxic content and spreading misinformation.
🚫 By simply converting harmful requests into past tense, many advanced LLMs "jailbreak," bypassing rejection training.
🔗 The solution is to include examples of past tense in training data to enhance the model's rejection capability for past tense requests.
Details Link: Research Paper
10. OpenAI Discusses Developing New Generation AI Server Chips with Broadcom and Others
OpenAI is in talks with Broadcom and other companies to develop new-generation AI server chips, aiming to enhance the company's computing power in AI development, reduce reliance on Nvidia GPUs, and strengthen market competitiveness.
AiBase Highlights:
🗣️ OpenAI CEO Sam Altman is leading the initiative to enhance the company's AI development computing power.
🤝 OpenAI has recruited former employees involved in Google TPU development to assist with chip design.
🚀 OpenAI's chip development plan aims to reduce reliance on Nvidia GPUs and increase competitiveness in the AI chip market.
11. Meta Pauses Generative AI Tool Training in Brazil
In Brazil, Meta's AI assistant was suddenly halted due to a ban by the National Data Protection Authority (ANPD). This decision has dampened Meta's plans to expand AI products in the Brazilian market and may also impact its global AI strategy.
AiBase Highlights:
🚫 ANPD prohibits Meta from using Brazilians' personal data to train AI models, imposing a daily fine of 50,000 reais.
💬 Meta has suspended the genAI features launched in Brazil and is in contact with ANPD to resolve the issue.
🔒 Data protection regulations pose challenges for global tech companies, and Meta must prioritize compliance and data privacy protection.
