Artificial intelligence

Generative models, agents, research, applications, and infrastructure for artificial intelligence.

38 Articles

Liquid AI Releases LFM2.5-2.6B Model to Run Agents Locally on Devices

Liquid AI announced the availability of the LFM2.5-2.6B and LFM2.5-2.6B-Base models on Hugging Face, with support for tool calling and multi-step workflows locally. The model targets running fast agents on computers and phones, while using less than 2.5 gigabytes of memory.

2026-08-07

Artificial intelligence

Cloudflare Enables AI Search to Build a Search Engine for AI Agent Data

Cloudflare announced improvements to AI Search that enable indexing files and websites, creating public endpoints through <strong>/search</strong> and <strong>/mcp</strong>, and connecting them to custom domains or securing them with Cloudflare Access. The company also revealed an experimental pricing model, while keeping the service free during the testing phase.

2026-08-06

Artificial intelligence

Cloudflare Launches Tools to Measure Website Readiness for AI Agents and Their Recommendation Rate

Cloudflare has added Agent Readiness and Answer Engine Optimization tools to its dashboard to help website owners determine how well AI agents can discover and read their content, and whether intelligent assistants recommend their websites and products. The tools include technical checks and metrics such as citation rate, mention rate, and share of visibility compared with competitors, with early access to AEO Visibility available.

2026-08-06

Artificial intelligence

Cloudflare: The Agentic Internet Needs Open Standards for Reading, Discovery, Execution, and Payment

Cloudflare views AI agents as a new type of web visitor, and says that treating them like traditional bots could mean blocking customers instead of blocking harmful traffic. The company proposes building an open agentic internet based on tools and standards that make content readable, discoverable, actionable, and payable.

2026-08-06

Artificial intelligence

TutorMoments Tests AI Models’ Ability to Know When to Help Students and When to Step Back

Ai2 presents the TutorMoments framework for evaluating language-model decisions during math tutoring sessions, focusing on the balance between providing support and encouraging students to think for themselves. Initial results showed that clarifying this balance in instructions improves model performance, but does not eliminate differences between models or close the gap with human teaching.

2026-08-06

Artificial intelligence

Cloudflare Launches Cloudflare OS as an Open Platform for Building AI Applications and Agents

Cloudflare announced the release of Cloudflare OS as an open-source project, giving organizations an environment that brings together AI agents, workspaces, and applications connected to internal systems. The platform focuses on access control and governance, with the ability to customize it and deploy it within the organization’s Cloudflare account.

2026-08-05

Artificial intelligence

Stack Overflow Launches a New Stack Internal Experience to Build a Trusted Knowledge Layer for AI

Stack Overflow announced new capabilities in Stack Internal that transform knowledge distributed throughout an organization into verifiable, usable context for teams and AI agents. The updates include connectors for Stack Internal Community, Google Docs, and Slack; a conversational interface that displays confidence levels and answer sources; as well as API and MCP integrations and permission-based access controls.

2026-07-30

Artificial intelligence

Microsoft Tracks Companies’ Shift from AI Experiments to “Frontier Transformation”

In its FY26 review, Microsoft examines how organizations have moved from testing artificial intelligence to integrating it into core operations, focusing on agents, systems governance, and the accumulation of knowledge and data within the enterprise. The company presents examples from multiple sectors, including financial services, healthcare, manufacturing, and retail, supported by customer-reported performance indicators.

2026-07-28

Artificial intelligence

Google Expands Managed Gemini Agents with Gemini 3.6 Flash and Environment Hooks

Google announced updates to Managed Agents in the Gemini API, including making Gemini 3.6 Flash the default model and adding environment hooks to control and audit tool calls. The update also introduces a free tier, budget limits, scheduled triggers, and an interface for managing isolated environment sessions.

2026-07-27

Artificial intelligence

Microsoft Invests $60 Million to Support Mission Genesis and Accelerate AI Research

Microsoft announced a long-term commitment to supporting the U.S. Department of Energy’s Mission Genesis through a $60 million investment and the launch of the SPARK center to coordinate collaboration among the company, national laboratories, and researchers.

2026-07-22

Artificial intelligence

FineBooks Tests the Ability of Open OCR Models to Recover Knowledge from Historical Books

A joint initiative between Hugging Face and EleutherAI evaluates the performance of 14 open OCR models on 2,165 historical pages and concludes that current models are generally suitable for improving language-model training data, but do not yet meet all the requirements of libraries or academic transcription.

2026-07-15

Artificial intelligence

Google expands Managed Agents capabilities in the Gemini API with background tasks and MCP server support

Google announced new updates for Managed Agents in the Gemini API, including running long-running tasks in the background, connecting to remote MCP servers, calling custom functions, and updating credentials between interactions. The updates are intended to help developers build agents better suited for operation in production environments.

2026-07-06

Artificial intelligence

Microsoft Launches Frontier Company with $2.5 Billion Investment to Expand AI Engineering

Microsoft announced the establishment of Microsoft Frontier Company, a new operating unit focused on helping customers implement AI-driven transformations, with a $2.5 billion investment and the involvement of 6,000 industry and engineering experts. The initiative focuses on delivering measurable business outcomes while protecting customers’ data and intellectual property.

2026-07-02

Artificial intelligence

Corporate Spending Data Suggests Anthropic’s Clash with the Trump Administration Could Boost Its Sales

Data from the Ramp platform shows that Anthropic’s share of companies’ paid AI subscriptions surpassed OpenAI’s in May, despite escalating tensions with the Trump administration and the withdrawal of its latest models from the market. An economist at Ramp believes the controversy could give the company an additional halo among customers, while its impact on a potential public offering remains unclear.

2026-06-16

Artificial intelligence

NVIDIA Showcases Rubin Platform and Open Self-Driving Models at CES 2026

During its keynote at CES 2026, NVIDIA unveiled the Rubin AI platform and the open Alpamayo family for developing autonomous vehicles, along with updates in personal computing, gaming and robotics.

2026-01-05

Artificial intelligence

UK-LLM Initiative Develops an AI Model Capable of Reasoning in Welsh

The UK-LLM initiative, in collaboration with University College London, Bangor University and NVIDIA, is working on an open AI model based on Nemotron that supports reasoning in Welsh and English. The model and its data are expected to be made available to businesses and the public sector, while Nscale will provide the model to developers through an API.

2025-09-13

Artificial intelligence

NVIDIA Research Outlines the Future of Physical AI

At SIGGRAPH 2025, NVIDIA unveiled software libraries, updates and models designed for physical AI, alongside the presentation of more than 12 research papers on 3D reconstruction, simulation and synthetic data. These developments target the training of robots and autonomous vehicles and the creation of more realistic virtual worlds.

2025-08-11

Artificial intelligence

Launch of Isambard-AI as the United Kingdom’s Most Powerful AI Supercomputer

The University of Bristol’s Isambard-AI supercomputer, powered by NVIDIA Grace Hopper chips, has entered service with a capacity of 21 exaflops for AI workloads. The system aims to accelerate research into drugs, climate, materials and language models, with access provided according to UK national priorities.

2025-07-17