OpenEuroLLM - Open-Source AI with European Standards
More topics: OpenAI presents an AI research assistant | GitHub introduces a new agent mode in VS Code
Hi AI Enthusiasts,
Welcome to this week’s Magic AI News, where we present you the most exciting AI and tech news of the week. Today, we’re talking about OpenEuroLLM, a family of open-source large language models that follow EU regulations.
In addition, we’re looking at OpenAI’s new “deep research“ feature in ChatGPT and GitHub’s new agent mode in VS Code! Stay curious! 😎
This week’s Magic AI tool is perfect for podcast fans. With this tool, you can instantly save insights and chat with episodes.
Let’s explore this week’s AI news together. 👇🏽
Top AI news of the week
💬 OpenEuroLLM: Open-Source AI with European Standards
A consortium of 20 leading European research institutions, companies, and EuroHPC aims to create a series of open-source foundation models for both private and public services.
The details:
- OpenEuroLLM aims to support the EU Commission’s goal of enhancing Europe’s competitiveness and digital sovereignty.
- Unlike other open-source models, the code of the models and the training data will be openly accessible. With other open-source models, you often don’t have access to the training data.
- OpenEuroLLM complies with EU regulations and is designed to offer a variety of transparent and powerful LLMs.
- According to the EU Commission, OpenEuroLLM currently has a total budget of €37.4 million.
Our thoughts
The amount of money sounds small compared to the 500 billion dollars of Project Stargate in the USA, but DeepSeek has shown that it is possible to create powerful LLMs with a smaller budget. After years of focusing on AI regulations, the EU Commission is trying to build its own models with European standards.
We think it’s good to have powerful open-source AI models that follow EU guidelines. That allows companies to use AI legally and continue to improve the models through an open-source approach. As Europeans, we are happy about this initiative.
More information: 🔗 OpenEuroLLM | European Commission
💻 OpenAI presents an AI research assistant
OpenAI has launched “deep research” as a new ChatGPT feature. This new functionality allows users to perform in-depth web research on complex topics and create detailed reports with citations.
The details:
- The system uses OpenAI’s o3 model, optimized for web browsing and data analysis. It can analyze text, images, and PDFs and provides a comprehensive research summary.
- You can select “deep research” in ChatGPT and enter a query. After the research started, a summary of all the steps appears in the sidebar.
- Deep research tasks can take between 5 to 30 minutes to complete. The system takes the time to dive deep into the web. You will be notified when the research is finished.
- OpenAI has developed “deep research” for people who do intensive knowledge work in fields like finance, science, or engineering.
Introduction video to Deep Research 👇🏽
Our thoughts
According to OpenAI, the new tool can create a comprehensive report at the level of a research analyst. Deep Research achieved 26.6% in Humanity’s Last Exam, significantly outperforming other AI models like OpenAI o1 (9.1%) and DeepSeek R1 (9.4%). Humanity’s Last Exam is a recently released multimodal benchmark at the frontier of human knowledge.
More AI systems are entering the market with enhanced reasoning abilities, leading to better results for complex tasks. However, these AI systems also require more computing time for inference because they perform in-depth research.
More information: 🔗 OpenAI | Humanity’s Last Exam
🤖 GitHub introduces a new agent mode in VS Code
GitHub announced several updates for GitHub Copilot and provides a first look at its software engineer (SWE) agent.
The details:
- The new agent mode of GitHub Copilot can independently review its code, detect errors, and fix them automatically. It suggests terminal commands to execute and analyzes runtime errors.
- Copilot Edits allows users in VS Code to make changes across multiple files using text commands. In addition, the system makes editing suggestions that are displayed inline in the editor.
- GitHub announced Project Padawan, an autonomous coding agent coming later this year. It can edit GitHub issues and create tested pull requests. Once a task is finished, Copilot assigns the pull request to a reviewer. So, Copilot becomes a contributor in GitHub repositories.
Here’s a short video that demonstrates how the new agent mode in VS Code works:
Our thoughts
Today, you can use AI coding tools for code completion or simple coding tasks. In the future, AI coding tools could evolve into coding partners and might even contribute independently to GitHub projects.
More information: 🔗 GitHub
Magic AI tool of the week
The tool Snipd allows you to mark highlights in a podcast. This way, you can access the most valuable information from the podcast later. Currently, the tool supports 11 languages (English, German, Spanish, French, …).
If you love podcasts, then try it out!
Step-by-Step Guide
- You hear something interesting.
- Triple-tap your headphones to save it to your library.
- An AI generates transcripts, summaries, and titles.
- Export your saved highlights to your favorite app like Notion*.
Articles of the week
- Our Journey to Become a Data Scientist
- Build an AI Finance Agent Team with phidata
- Understand and Implement an Artificial Neural Network from Scratch
- Visualize financial data with beautiful Sankey diagrams
💡 Do you enjoy our content and want to read super-detailed articles about AI? If so, subscribe to our blog and get our popular data science cheat sheets for FREE.
Thanks for reading, and see you next time.
- Tinz Twins
