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
After a six-month beta period and more than 500,000 interviews, HackerRank is making Chakra available to all its clients. Chakra is an agent that uses AI to observe candidates as they code and asks them about their choices.
The AI agent answers questions about products, sizes or delivery and remembers the user’s preferences, whilst Buy Direct allows you to shop with a brand without leaving the video feed.
The New York-based start-up founded by former DeepMind employees claims to match GLM-5.2 and come close to Qwen 3.8-Max in terms of code and agent-based tasks, whilst using three to four times less computational power.
From the end of October, OpenAI will be trialling an illustrated advertising format in the US – displayed whilst ChatGPT generates an image – and is enhancing its analytics tools for advertisers.
Trained on 90,000 real queries from researchers, Ai2’s small model generates a cited report in a single pass and powers the new ‘Fast’ mode on the Asta platform.
A lighter version of the DGX Spark mini-supercomputer arrives this month from six manufacturers, with the same GB10 chip as the 128 GB model and the possibility of pairing two of them.
Until now reserved for paying subscribers, the tool that links bank accounts (via Plaid) and credit report (via Experian) is coming to all American users.
Clef (27 billion parameters) and Clef-flash (9 billion) do not generate text: they return typed answers with probabilities, in open source under the Apache 2.0 license.
A selfie or a full-length photo is enough to view a piece of clothing or an accessory on you, thanks to the Images 2.5 model. OpenAI also adds a Favorites function.
Ideogram's new model limits pixel, color, and texture drift from one edit to another, and retouches high-resolution images without shrinking them.
The Paris-based laboratory has released, under the MIT licence, a multilingual speech synthesis model – including French – which is six times faster than real time on a MacBook Air.
The new model positions each element on a coordinate grid, edits a specific area without affecting the rest, and combines up to 10 reference images.
Currently available as a public preview in Copilot CLI and the Copilot app, this feature reads, clicks and types within software that lacks an API, following the user’s explicit consent.
The result of a multi-year agreement, GPT-Synopsys is trained to handle Synopsys' EDA software like an engineer, to iterate on the consumption, performance and surface area of the chips.
The real-time transcription model produces its first hypotheses in a hundred milliseconds in 60 languages; the Flash variant of text-to-speech targets low-latency agents.
Fermion Research has released an English speech recognition model that takes up 164 MB, has an error rate of 5.21 per cent per word and transcribes an hour of audio in around 20 seconds on a MacBook Air.
Released under the MIT licence, the library implements the DeepGEMM API for Ascend 950 NPUs in BF16, FP8 and FP4 — another step towards phasing out CUDA in China.
Cohere’s two new embedding models share the same vector space, process 128,000 tokens, support over 100 languages and images, and cost $0.12 and $0.08 per million tokens respectively.
Launched in beta in July, Anthropic’s offering for federal and state government agencies is now available in a FedRAMP High-certified environment.
As the first model in the Gemini 4 generation, Argon is designed for coding, business use and cyber defence. It will first be made available through the Fairwind programme before being rolled out to Ultra subscribers and via the API.