AM-IA brings together news, research papers and applications relating to artificial intelligence published around the world into a single feed — and makes them available in your European language at the click of a button.
Scroll to browse · Click to filter
The contact centre specialist has introduced a ‘long-term’ agent capable of putting a case on hold whilst awaiting a document, and then resuming it independently, subject to human approval.
Two weeks after a launch restricted to cybersecurity professionals, Google is making its Edge model available to developers, businesses and Pro and Ultra subscribers.
Following on from Instagram, Meta is allowing Facebook users to specify the topics they want to see more or less of, with recommendations that adjust in real time.
With just a Google account, you can describe an app to the AI and install it on your phone with a single click, via a private section of Google Play, without going through the public Play Store.
With the number of code merges rising from 2,500 to over 30,000 per day between May and September, Anthropic migrated its monorepo from Git to Scion, an in-house tool designed for agents that are always online.
The former SpaceXAI is launching a standardised pricing structure across its products: four tiers provide access, to varying degrees, to the Grok models, the Cursor code editor (acquired in August) and the paid features of X.
Trained on the Prometheus data centre, the two models rely on active memory and greater resilience in the face of obstacles. They are available exclusively to paying subscribers and developers.
A model with 7 billion parameters that speaks the Emirati dialect, a transcription model for six languages and an Arabic text reader: the Emirati institute is focusing on languages that are underserved by large language models.
Maia 300, Microsoft’s third-generation AI chip, promises 40 per cent more tokens per megawatt than GPU-based systems and will be deployed on a gigawatt scale from 2027.
Rather than an orchestrator that delegates tasks to specialised models, Reka presents a single network that comprehends, generates and acts, with a distilled version capable of rendering 5 seconds of video in approximately one second.
A side panel allows Claude to view the open file and edit it on the spot, with each change saved by default; connectors also create Google files directly from the chat.
Aleph Alpha’s first open-source model is a mixture-of-experts model with 78 billion parameters – 3.46 billion of which are active per token – focused on German and English, with a context validated up to one million tokens.
Derived from Qwen3.5-9B and available in Foundry and on OpenRouter, the model assigns a calibrated probability to each option for routing, classifying or verifying within agent workflows.
Rather than generating text token by token, these ‘decision models’ return a short, structured response in a single inference, for real-time embedded applications.
Hundreds of games and emulators, reconstructed using AI-assisted decompilation, have appeared in the space of a few weeks. Kotaku has tested web ports of classics that work almost perfectly.
It has the same architecture as Qwen-Image-2.1, with 7 billion parameters, but only eight denoising stages: the checkpoint is available on Hugging Face under a research licence.
A week after Clef, Cloudflare has launched Clef-omni, which can make decisions based on text, images, audio and video in a single call, and has reduced the price of Clef-flash.
Ahead of Argon’s public roll-out, Google’s teams are testing a new version of Gemini 4 on their internal code platform. Its release is not guaranteed.
Backed by Khosla Ventures, the start-up is developing a non-invasive cap fitted with 100,000 sensors, coupled with its own brain model, to control AI agents without the need for a keyboard or voice commands.
Thanks to Claude Managed Agents, AI agents will be able to join meetings, calls and chat rooms to continue the work once the team has signed off.