Spend 1 minute every day to get selected cutting-edge AI information.

The content includes but is not limited to cutting-edge AI news, AI tools, AI painting, open source projects, and learning tutorials, etc.

Follow AI Daily to keep up with the AI trend, and hope it helps you. For important information, separate posts will be made for detailed introductions.

Here is the latest AI information for July 26.

Cutting-edge News

1. OpenAI is about to launch the AI search product SearchGPT!

Aiming to provide us with fast, timely, and clear answers with relevant sources by combining the powerful capabilities of AI large language models and real-time information on the internet.

Detailed introduction: https://openai.com/index/searchgpt-prototype/

Currently, only a temporary prototype is provided, and it is being beta tested by a small number of users. If you are interested, you can apply to join the waitlist.

Can it threaten the most powerful AI search tool Perplexity? Let's wait and see!

2. Mistral AI releases the Mistral Large 2 model.

With a 128K context window and 123B parameters, its capabilities in code generation, math, and reasoning are significantly enhanced, while also providing stronger language support and advanced function call capabilities.

In benchmark tests, its overall performance is comparable to models like GPT-4o, Claude 3 Opus, and Llama 3 405B.

Detailed introduction: https://mistral.ai/news/mistral-large-2407/

Model download: https://huggingface.co/mistralai/Mistral-Large-Instruct-2407

Online experience: https://chat.mistral.ai/chat

3. Another image-generated 3D model Aiuni.

Developed by a post-00s team from Tsinghua and Peking University, the product's effect is very good, with good detail handling, and can be directly experienced online.

Official website: https://aiuni.ai/#/

The following image was generated by DALL-E and then converted to 3D effect.

Cutting-edge Technology

1. A high-quality, human-like simultaneous interpretation system CLASI developed by ByteDance.

It adopts a new read-write strategy to balance translation quality and delay, and enhances translation quality through multimodal large language models for error correction.

Detailed introduction: https://byteresearchcla.github.io/clasi/

According to their tests, it can provide accurate translation results fluently and naturally in scenarios such as fast speech, tongue twisters, and impromptu conversations.

Currently, only the test dataset is open, and it is not yet available for experience. Interested parties can first watch the provided demo video, which looks quite good.

Open Source Projects

1. An open-source, free AI personal tutor tool Llama Tutor on GitHub.

Built with Next.js and Tailwind CSS, it integrates Llama 3.1 and Serper to provide strong reasoning and search capabilities for high-quality auxiliary teaching.

GitHub: https://github.com/Nutlope/llamatutor

Online experience: https://llamatutor.together.ai/

Enter any knowledge you want to know, and choose the acceptable educational level, covering from elementary school to university, to generate a personalized tutor that answers all your questions.

2. A TV live streaming software developed natively for Android: My TV.

It supports customizable live sources, supports m3u and tvbox formats, and allows setting multiple live sources, supports customizable program lists, and automatic switching of multiple lines.

GitHub: https://github.com/yaoxieyoulei/mytv-android

If you have a need, you can download and try it.