Each week, Quartz brings together product launches, updates, and funding news for AI-focused startups and companies.

Here’s what’s happening this week in the ever-evolving AI industry.

Cohere’s multilingual model family, Aya Expanse

Cohere For AI, the research arm of enterprise AI platform Cohere, this week announced its family of multilingual models, Aya Expanse.

The family of “high performance” models works in 23 languages ​​and has been released as both an 8 billion parameter model and a 32 billion parameter model on Kaggle and Hugging Face. The parameters are variables that a model learns from training data that guides its ability to make predictions.

The smaller model “makes discoveries more accessible to researchers around the world,” while Cohere’s larger model “offers industry-leading multilingual capabilities,” the company said. Multilingual models outperformed Google’s open models (GOOGLE), Mistral and Méta (META), according to Cohere.

Gefion, Denmark’s first AI supercomputer

NVIDIA (NVDA) CEO Jensen Huang and King Frederik X of Denmark unveiled the the largest sovereign AI supercomputer, Gefionthis week alongside Nadia Carlsten, Managing Director of the Danish Center for AI Innovation. The country’s first AI supercomputer is named after a goddess from Danish mythology and is operated by the Danish Center for AI Innovation. The AI ​​supercomputer is an Nvidia DGX SuperPOD powered by 1,528 Nvidia H100 Tensor Cores graphics processing units (GPUs).

Gefion is still in the works and will launch in a pilot phase for projects focused on using “AI to accelerate progress, including in areas such as quantum computing, drug discovery and energy efficiency “, said Nvidia.

Amazon One Medical’s new AI tools

Amazon One Medical (AMZN) launched a new AI tools for its healthcare providers this week to reduce time spent on administrative tasks such as reviewing medical records and taking notes. The new AI tools are integrated into Amazon One Medical’s proprietary electronic health record.

AWS HealthScribe takes notes of conversations during patient visits in real time, allowing providers to engage more with patients, Amazon said. Then, providers can review and update the scores before submitting them.

Another new AI tool can read, label and summarize medical records from outside Amazon One Medical, allowing providers to search for screenings, results and medication information.

Luminance’s legal AI agent, Agent Lumi

Luminance launched Agent Lumi this week, which combines the company’s proprietary and area-specific legal knowledge large language model (LLM)AI tools and memory. Agent Lumi “performs tasks and makes decisions without the need for human oversight,” the company said.

Active AI agents can automate legal tasks on top of Luminance’s LLM, which was trained on over 150 million verified legal documents. The Lumi Agent also has short- and long-term memory, meaning it can “retain and recall information, learning from past interactions with individual users to adapt its actions and responses over time.” time,” according to the company.

Legal AI Platform Alexi’s all-in-one platform

AI litigation platform Alexi this week announced an all-in-one platform for automate up to 80% of common legal tasksgiving litigators more time for “high-impact work,” the company said.

The upgraded Alexi platform includes new features such as an AI conversational assistant and the ability to analyze and summarize legal documents of any length or format.

“This new Alexi product builds on our strong foundation in AI-powered legal research products; Today, however, we are revolutionizing the entire litigation workflow,” said Mark Doble, Managing Director of Alexi.

Clinical decision support platform, Wound AI

Real-time intelligent applications platform Vantiq, Telemedicine Solutions and NTT DATA launched Wound AI this week, which they call a “state-of-the-art clinical decision support platform” for wound treatments. The platform uses generative AI to recommend precise wound care and treatment selection assistance.

Wound AI can integrate with existing electronic health records “and is built on a federated AI model that prioritizes data privacy and security while promoting learning and continuous improvement,” they said. declared the companies.

For the latest news, Facebook, Twitter And Instagram.