Welcome to the 【AI Daily】 column! Here is your daily guide to exploring the world of artificial intelligence. Every day, we bring you the latest hot content in the AI field, focusing on developers and helping you gain insight into technological trends and understand innovative AI product applications.
Fresh AI products click to learn more: https://top.aibase.com/
1. Cursor Pro removes the 500 request limit, opening a new chapter of unlimited use
As a developer, I am very happy to see that Cursor Pro has removed the monthly 500 quick request limit. This undoubtedly greatly improves development efficiency. However, I also pay attention to the potential impact of rate limits and suggest everyone to plan their usage reasonably based on their own needs.
【AiBase Summary:】
🎉 Cursor Pro plans to remove the 500 quick request limit and introduce the 'unlimited use' mode.
💼 Launches the new Ultra plan at $200 per month, offering model usage equivalent to 20 times the Pro plan.
📊 Anysphere valuation reaches $9.9 billion with annual revenue exceeding $500 million, consolidating its market leadership position.
2. Xiyu Technology's MiniMax launches video generation model Hailuo 02
Xiyu Technology launches its new video generation model Hailuo02, which performs excellently in handling complex scenes and improving creative quality, significantly reducing the threshold for video creation.
【AiBase Summary:】
✨ Hailuo02 is the only model capable of efficiently handling complex scenarios like gymnastics movements, already helping creators generate over 370 million videos.
⚙️ NCR architecture enhances model training and inference efficiency by 2.5 times, increasing parameter size threefold and supporting larger-scale data fitting.
🎥 Supports 1080p video generation with high cost-effectiveness, available on multiple platforms with various resolution options.
Details link: https://hailuoai.com/create
3. Google releases Gemini 2.5 Flash-Lite: Faster inference speed, lower costs!
Google introduces Gemini 2.5 Flash-Lite, the lightest and most cost-effective AI model in its series, featuring fast inference and low latency, suitable for various applications including real-time translation and high-throughput classification tasks.
【AiBase Summary:】
⚡️ Gemini 2.5 Flash-Lite has fast inference speed and low latency, suitable for real-time translation and high-throughput classification tasks.
🔍 Supports ultra-long context processing up to 1 million tokens, with strong flexibility for complex system development.
🌐 Integrated into Google AI Studio, Vertex AI platform, and Google Search, assisting developers in efficient work.
4. iFlytek Strikes Again! New Starfire X1 Upgrade Version Set to Launch in July!
iFlytek is about to release the upgraded version of Starfire X1, expected to launch in July. This version shows significant improvements in performance, user experience, and market layout, demonstrating iFlytek's continuous innovation capability in the intelligent voice field.
【AiBase Summary:】
🌟 The upgraded version of Starfire X1 will be released in July, with overall performance enhancement and providing a smoother and smarter user experience.
🌐 iFlytek collaborates with multiple enterprises to accelerate market expansion, highlighting its leading position in the digital economy.
💻 Significant improvement in user interface and interaction experience, bringing more efficient and convenient application experience.
5. Tencent Yuanbao Launches AI Coding Mode: Real-time Code Generation and Preview
Tencent Yuanbao launches a new AI coding mode, generating code in real time through a double-column interface and previewing the results, supporting multi-language online execution, reducing learning barriers, especially suitable for educational and family scenes.
【AiBase Summary:】
🌟 Users can generate and preview code in real time through a double-column interface, enhancing development efficiency.
🚀 Supports online execution of multiple programming languages without environment configuration, simplifying operational processes.
👨👩👧👦 Suitable for educational and family scenes, stimulating creativity and programming interest.
6. OpenAI Announces GPT-4.5 Will Be Discontinued from API, Developers Feel Confused and Disappointed
OpenAI plans to discontinue the GPT-4.5 preview version from the API on July 14, 2025, causing significant impact on developers who depend on this model. Nevertheless, GPT-4.5 remains available for ChatGPT personal users, and OpenAI recommends developers transition to GPT-4.1 or other models.
【AiBase Summary:】
📅 OpenAI will discontinue the GPT-4.5 preview version from the API on July 14, 2025, a major change developers need to adapt to quickly.
🤖 Although GPT-4.5 no longer supports API usage, it remains a tool for ChatGPT personal users, providing continuous service.
🔄 OpenAI encourages developers to migrate to GPT-4.1 or other alternative models to ensure business continuity and reduce operating costs.
7. Apple’s New Speech API Transcription Speed Surpasses OpenAI Whisper by 55%
Apple’s newly launched Speech API demonstrates excellent transcription speed, completing the transcription of a 34-minute 4K video in just 45 seconds, far surpassing other tools like OpenAI’s Whisper. The advantage of this technology lies not only in speed but also in its local computing capability, making multi-segment video processing more efficient.
【AiBase Summary:】
🌟 Apple’s new Speech API transcribes a 34-minute 4K video in just 45 seconds, surpassing competitors.
⏱️ Compared to OpenAI Whisper, Apple’s technology improves efficiency by approximately 55%, showing significant improvement.
📈 The advantage of local computation makes Yap more efficient in multi-segment video processing, saving users a lot of time.
8. Baidu Introduces Global First Dual Digital Human Interactive Live Stream Driven by Wenxin Large Model 4.5T
This article introduces Baidu’s global first dual digital human interactive live stream, driven by Wenxin large model 4.5T, showcasing breakthroughs in multimodal technology, having a profound impact on the live streaming industry and content creation ecosystem.
【AiBase Summary:】
🌟 Wenxin large model 4.5T achieves multimodal joint modeling, significantly enhancing digital human interaction experiences.
🌐 Dual digital human live streams assist e-commerce, education, and other fields in content creation, reducing costs and enhancing diversity.
🤝 Open-source plans promote the popularization of multimodal AI technology, assisting small and medium-sized enterprises and developers in innovative applications.
9. OpenAI CEO Exposes That Meta Once Tried to Poach OpenAI Talent for $100 Million but Failed
Meta once offered high salaries to poach top AI researchers from OpenAI and Google DeepMind but failed. OpenAI CEO Sam Altman believes that OpenAI’s innovative culture and AGI development prospects attract excellent employees. Meanwhile, OpenAI is developing an AI-based social media application to compete with Meta.
【AiBase Summary:】
📌 Meta once made a high offer to poach talent from OpenAI but failed to successfully recruit excellent employees.
💼 Sam Altman believes that OpenAI employees value the company’s innovative culture and AGI development prospects more.
🤖 OpenAI is developing an AI-based social media application, attempting to compete with Meta in the social network domain.
10. Say Goodbye to "AI Sense"! Krea1 Beta Test Open, Super-realistic Textures and Details, Diverse Art Styles
Krea AI’s first image generation model, Krea1 beta version, is now open for testing, solving the traditional AI image generation problem of "AI sense," providing super-realistic textures, diverse styles, and personalized customization support, along with real-time editing functions.
【AiBase Summary:】
✨ Super-realistic textures and details: Support 1.5K native resolution, up to 4K ultra-high definition, showcasing delicate textures and realistic materials.
🎨 Diverse art styles: Accurately understands complex art style requirements, avoiding the monotony of "AI images."
🌟 Personalized customization support: Allows uploading reference images or datasets, real-time adjustment of generated results, enhancing creator control.
11. Tesla’s Grok In-Car AI Assistant Set to Launch, Personalization Makes Driving More Fun
Tesla’s Grok in-car AI assistant is expected to bring owners a new smart experience, featuring interactivity and personalization, including different ‘personality’ versions and child mode, but currently limited to AMD chip models. As the Robotaxi platform advances, the importance of Grok becomes increasingly prominent.
【AiBase Summary:】
🚗 Grok in-car AI assistant will break the traditional boundaries between people and cars, providing rich communication methods such as answering questions and personalization.
👨👩👧👦 Supports various personalization options, such as different personality versions and child modes, suitable for family users.
🔄 Limited to AMD chip models, new models will enjoy stronger smart technologies, promoting the maturity of in-car AI technology.
12. Google Gemini Gets Major Upgrades! Video Upload and Analysis Functions Officially Launched
Google officially launches Gemini 2.5 Pro and Flash versions, fully available to users, and adds video upload and analysis functions, supporting Android and web platforms, enhancing video processing capabilities.
【AiBase Summary:】
🌟 Gemini adds video upload and analysis functions, allowing easy upload of videos for content analysis on Android and web platforms.
🎥 Gemini can comprehensively analyze video content, including overview, finding clips or items, and displaying related segments, enhancing video processing efficiency.
🏆 In the video analysis field, Gemini surpasses ChatGPT, providing a more convenient user experience and consolidating market competitiveness.