InfoQ Homepage Large language models Content on InfoQ
-
Hugging Face to Democratize Robotics with Open-Source Reachy 2 Robot
Hugging Face has acquired Pollen Robotics, a French startup that developed the humanoid robot Reachy 2. The acquisition aims to make robotics more accessible by open-sourcing the robot’s design and allowing developers to modify and improve its code.
-
Meta Launches AutoPatchBench to Evaluate LLM Agents on Security Fixes
AutoPatchBench is a standardized benchmark designed to help researchers and developers evaluate and compare how effectively LLM agents can automatically patch security vulnerabilities in C/C++ native code.
-
OpenAI Launches BrowseComp to Benchmark AI Agents' Web Search and Deep Research Skills
OpenAI has released BrowseComp, a new benchmark designed to test AI agents' ability to locate difficult-to-find information on the web. The benchmark contains 1,266 challenging problems that require agents to persistently navigate through multiple websites to retrieve entangled information.
-
Cloudflare AutoRAG Streamlines Retrieval-Augmented Generation
Cloudflare has launched a managed service for using retrieval-augmented generation in LLM-based systems. Now in beta, CloudFlare AutoRAG aims to make it easier for developers to build pipelines that integrate rich context data into LLMs.
-
Scaling Financial Operations: Uber’s GenAI-Powered Approach to Invoice Automation
Uber recently described a GenAI-powered invoice processing system that reduced manual effort by 2x, cut handling time by 70%, and delivered 25–30% cost savings. By leveraging GPT-4 and a modular platform called TextSense, Uber improved data accuracy by 90%, enabling globally scalable, efficient, and highly automated financial operations.
-
Docker Bridges Agents and Containers with New MCP Catalog and Toolkit
Docker has announced two new AI-focused tools—the Docker MCP Catalog and the Docker MCP Toolkit—to bring container-grade security and developer-friendly workflows to agentic applications, helping build a developer-centric ecosystem for Model Context Protocol (MCP) tools.
-
Google's Gemma 3 QAT Language Models Can Run Locally on Consumer-Grade GPUs
Google released the Gemma 3 QAT family, quantized versions of their open-weight Gemma 3 language models. The models use Quantization-Aware Training (QAT) to maintain high accuracy when the weights are quantized from 16 to 4 bits.
-
Google DeepMind Shares Approach to AGI Safety and Security
Google DeepMind has released a new paper outlining its approach to safety and security in the development of artificial general intelligence (AGI). AGI refers to AI systems that are as capable as humans at most cognitive tasks.
-
PayPal's New Agent Toolkit Connects AI Frameworks with Payment APIs through MCP
PayPal has released its Agent Toolkit, designed to help developers integrate PayPal's API suite with AI frameworks through the Model Context Protocol (MCP). The toolkit provides access to APIs for payments, invoices, disputes, shipment tracking, catalog management, subscriptions, and analytics capabilities.
-
AWS Promotes Responsible AI in the Well-Architected Generative AI Lens
AWS announced the availability of the new Well-Architected Generative AI Lens, focused on providing best practices for designing and operating generative AI workloads. The lens is aimed at organizations delivering robust and cost-effective generative AI solutions on AWS. The document offers cloud-agnostic best practices, implementation guidance and links to additional resources.
-
DeepMind Researchers Propose Defense against LLM Prompt Injection
To prevent prompt injection attacks when working with untrusted sources, Google DeepMind researchers have proposed CaMeL, a defense layer around LLMs that blocks malicious inputs by extracting the control and data flows from the query. According to their results, CaMeL can neutralize 67% of attacks in the AgentDojo security benchmark.
-
Microsoft Native 1-Bit LLM Could Bring Efficient genAI to Everyday CPUs
In a recent paper, Microsoft researchers described BitNet b1.58 2B4T, the first LLM to be natively trained using "1-bit" (technically, 1-trit) weights, rather than being quantized from a model trained with floating point weights. According to Microsoft, the model delivers performance comparable to full-precision LLMs of similar size at a fraction of the computation cost and hardware requirements.
-
Google DeepMind Introduces QuestBench to Evaluate LLMs in Solving Logic and Math Problems
Google DeepMind’s QuestBench benchmark helps in evaluating if LLMs can pinpoint the single, crucial question needed to solve logic, planning, or math problems. DeepMind team recently published an article on QuestBench which is a set of underspecified reasoning tasks solvable by asking at most one question.
-
Docker Model Runner Aims to Make it Easier to Run LLM Models Locally
Currently in preview with Docker Desktop 4.40 for macOS on Apple Silicon, Docker Model Runner allows developers to run models locally and iterate on application code using the local models- without disrupting their container-based workflows.
-
AI Continent: European Commission Outlines Strategy for Scaling AI Development
The European Commission has presented the AI Continent Action Plan, a new strategy designed to strengthen the European Union’s capacity for AI development and deployment. The plan outlines coordinated investment in infrastructure, access to high-quality data, AI adoption in strategic sectors, and support for regulatory implementation.