Grok 3 Beta achieves SOTA results in finance benchmarks
The US-based company Vals AI specializes in evaluating and benchmarking large language models (LLMs) for specific industry use cases. The goal is to measure ...
Magic AI
Posts by
Years TagsRecent posts
EU plans five AI gigafactories
The EU aims to strengthen Europe’s AI sector and plans to establish five AI gigafactories starting in 2026.
Firebase Studio - Build full-stack apps with Gemini
Google released a cloud-based, agentic development environment powered by Gemini called Firebase Studio. With Firebase Studio, you can build full-stack AI ap...
Llama 4 doesn’t meet the expectations of the AI community
A week ago, we talked about the release of Llama 4. Now, the first critical voices are coming from the AI world. And this is justified for several reasons.
Llama 4 - A Multimodal Model Family with MoE Architecture
Meta just released the first models of the Llama 4 model family. The new models offer unmatched speed and efficiency. Meta calls it “the beginning of a new e...
Amazon Nova Act - An SDK for web agents
Amazon AGI Labs introduced a new AI model designed to perform actions in a web browser. In addition, they released an AI agent framework called Nova Act SDK ...
GPT-4.5 achieves 73% in Turing test
Researchers at UC San Diego have demonstrated that current large language models can pass the famous Turing test.
Docker Model Runner - Run AI Models Locally
Docker announced AI Model Runner for Docker Desktop. This feature is in preview for Docker Early Adopters. It could be a game-changer for AI developers.
ElevenLabs releases Actor Mode
The voice technology startup ElevenLabs* just launched Actor Mode. It is now available in ElevenLabs Studio, an advanced speech editor.
ChatGPT has a new image generator
OpenAI has released image generation using the GPT-4o model in ChatGPT, moving from separate text and image models to an integrated approach. GPT-4o replaced...
NVIDIA unveils AI and robotics updates at GTC 2025
At the GTC 2025 conference in San Jose, CEO Jensen Huang highlighted the company’s latest advancements in AI, robotics, and computing.
Open-Sora 2.0 - Open-source video AI with low training costs
The Singapore-based AI startup HPC-AI Tech has introduced a video AI model with lower training costs compared to other similar models. The startup called the...
Mistral AI presents a new super-fast AI model
Mistral AI announced Mistral Small 3.1. It’s an open-source model that outperforms similar models like GPT-4o Mini or Gemma 3 in several benchmarks and terms...
Gemma 3 - An AI model that can run on a single GPU
Google has introduced the next generation of its open-source models called Gemma 3. It is a family of lightweight open models built with the same technology ...
Manus AI presents an AI agent system
The Chinese AI startup Manus AI has introduced a preview of its AI agent system, called Manus. The AI agent can assist with travel planning and financial ana...
Gibberlink - Communication between two AI agents
During a hackathon project in London, the developers Anton Pidkuiko and Boris Starkov developed a method that enables AI agents to communicate more efficient...
Grok now replies to X users
X users can now publicly ask Grok, the AI chatbot of xAI, questions by mentioning the AI. Then, Grok answers the questions automatically. In addition, xAI in...
Mistral OCR understands complex documents
The French AI startup Mistral introduced an accurate document understanding API, called Mistral OCR. OCR stands for Optical Character Recognition.
Deutsche Telekom announces an AI Phone
Deutsche Telekom, the parent company of T-Mobile, announced an AI Phone in partnership with Perplexity at the Mobile World Congress in Barcelona.
OpenAI presents GPT-4.5
OpenAI released a research preview of GPT-4.5 (code-named Orion). It uses unsupervised learning to achieve deeper world knowledge. According to OpenAI, it’s ...
ElevenLabs releases a new speech-to-text model
ElevenLabs launched their first speech-to-text model called Scribe. It can transcribe speech in 99 languages.
Inception Labs presents diffusion large language models
The AI startup Inception Labs announced a new generation of large language models (LLMs) with the Mercury family. The Mercury models are diffusion LLMs (dLLM...
Fear of AI in professions varies by country
A study in the American Psychologist shows that people’s fear of AI varies by profession and country.
Google presents a multi-agent AI co-scientist
Google introduced AI co-scientist, a multi-agent AI research assistant built with Gemini 2.0, to help scientists generate and verify new ideas in fields like...
xAI unveils Grok 3 Beta
xAI introduced Grok 3, a family of advanced models (Grok 3 and Grok 3 mini) with strong reasoning capabilities and extensive pretraining knowledge. Before th...