What is today’s beat?

RELEASES
🧨 EU LAUNCHES AN OPEN SOURCE CONSORTIUM
🧨 IMAGEN 3 NOW AVAILABLE THROUGH GEMINI API

NEWSROOM
🗞️ ARX Robotics opens Europe's largest defense robotics facility
🗞️ Barcelona's KOA Biotech secures €2M to tackle aquaculture challenges
🗞️ Emblem raises $85M for its initial fund
🗞️ Lyft partners with Anthropic to enhance customer service with AI

BUILDER BYTES
⭐️ Advances to low-bit quantization enable LLMs on edge devices
⭐️ Studio is now available to everyone
⭐️ mistralai/Mistral-Small-24B-Instruct-2501

COMMUNITY
🤩 Agent Engineering with Pydantic + Graphs
🤩 Vertical AI Agents Could Be 10X SaaS | YCombinator
🤩 DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters | Lex Fridman Podcast #459
🤩 Andrej Karpathy, Deep Dive into LLMs like ChatGPT

FOLLOW US ON SOCIAL FOR MORE

Your FREE newsletter
 share or subscribe 
to show support

🧨 EUROPEAN UNION 🧨

EU LAUNCHES AN OPEN SOURCE CONSORTIUM

A consortium of 20 European research institutions, companies, and EuroHPC centers has initiated the OpenEuroLLM project to develop open-source, multilingual large language models (LLMs). Coordinated by Jan Hajič of Charles University and co-led by Peter Sarlin of AMD Silo AI, the project aims to create transparent and compliant AI models for diverse applications.

• Multilingual Open-Source LLMs
  The project focuses on building high-performance, multilingual LLMs that are open-source, ensuring transparency and adaptability for various sectors.

• Alignment with European Regulations
  Developed within Europe's regulatory framework, these models will adhere to European values and standards, promoting responsible AI development.

• Collaboration with Open-Source Communities
  By partnering with organizations like LAION and OpenML, the consortium ensures that models, software, and data are fully open and customizable for specific industry and public sector needs.

• Preservation of Linguistic Diversity
  The initiative emphasizes supporting linguistic and cultural diversity, enabling the development of AI products and services across various European languages.

The project seeks to align with Europe’s regulatory framework while ensuring that AI development remains accessible and adaptable to various needs. By collaborating with organizations such as LAION, OpenML, and open-sci, OpenEuroLLM plans to release models that support linguistic diversity and can be fine-tuned for specific industry and government use cases.

🧨 GOOGLE 🧨

IMAGEN 3 NOW AVAILABLE THROUGH GEMINI API

Google has integrated its advanced image generation model, Imagen 3, into the Gemini API, enabling developers to create high-quality images across various styles. Initially available to paid users, access will soon extend to the free tier.

• Diverse Image Generation
  magen 3 produces artifact-free images in styles ranging from hyperrealism to abstract art and anime.

• Enhanced Prompt Adherence
  The model demonstrates improved accuracy in interpreting textual prompts, facilitating the creation of images that closely match user specifications.

• Cost-Effective Access
  Priced at $0.03 per image, developers can control aspects like image count and aspect ratios, making it a flexible tool for various project needs.

• Digital Watermarking
  To combat misinformation, all images generated include a non-visible SynthID watermark, identifying them as AI-generated.

Check out Google’s recent blog with code samples, or their image prompt guide for more details on how to use it.

⭐️ AI Leaderboard ⭐️

Quick glance at the leading AI models this week

• 🤩 Top Performers ⭐️ 
  OpenAI leads with o3-mini-2025-01-31-high (75.88) and o1-2024-12-17-high (75.67) lfor leading global average rankings.
  

• 💪 Strength & Utility 🔍️
  OpenAI dominates reasoning
  DeepSeek's deepsseek-r1 excels in coding
  Google for integrated functionality (IF).
  

• 💥 Google & Alibaba 💥
  Google's gemini models have mixed performance, with strong IF scores but weaker language and coding scores; Alibaba's qwen2.5-max ranks mid-tier.

🗞️ NEWSROOM 🗞️ 

What’s hot in tech right now?

ARX Robotics opens Europe's largest defense robotics facility

ARX Robotics has inaugurated a new facility in Munich, significantly boosting its capacity to produce autonomous military robots at scale. This expansion aims to enhance Europe's defense capabilities with advanced unmanned ground systems.

Barcelona's KOA Biotech secures €2M to tackle aquaculture challenges

KOA Biotech has raised €2 million to advance its biosensor technology for early detection of waterborne infections in fish farms. The innovation seeks to reduce fish mortality and antibiotic use in aquaculture.

Emblem raises $85M for its initial fund

Emblem, a European VC firm, has closed its inaugural fund at €80M ($85M), aiming to invest in early-stage startups across sectors to boost European tech innovation.

Lyft partners with Anthropic to enhance customer service with AI

Lyft is integrating Anthropic’s Claude AI into customer support, cutting resolution times by 87%. AI handles common queries while human agents tackle complex issues.

⭐️ BUILDER BYTES ⭐️

What’s hot for builders right now?

Advances to low-bit quantization enable LLMs on edge devices
Discusses how low-bit quantization techniques, such as T-MAC and Ladder, improve the efficiency of running large language models on edge devices by reducing computational demands and enhancing hardware compatibility.

Studio is now available to everyone
Announces that Studio, a longform text-to-audio editor for creators and storytellers, is now accessible to all users, offering features like multiple character voices, fine-grained voice customization, and pacing control.

mistralai/Mistral-Small-24B-Instruct-2501
Trending in #4 position on HuggingFace, Mistral AI’s model is gaining popularity and has a recent update just a few days ago.

🤩 COMMUNITY 🤩

What’s the latest beat?

THANK YOU

Looking to promote your company, product, service, or event?