Companies

The AI startup released migration instructions for OpenClaw users after Anthropic restricted model access on open agent platforms.

The open-source post-training framework used to fine-tune and align large language models reaches its first major stable release.

Technology Innovation Institute expands open-source Falcon family with a 600-million-parameter multimodal model for object detection and segmentation.

Google has released Gemma 4, a frontier multimodal model designed to bring advanced vision-language capabilities to edge hardware.

The widely adopted secure tensor serialization format will now be governed by the PyTorch Foundation, marking a step toward industry standardization.

The open-source machine learning platform expanded its popular embedding library to support combined text and image retrieval.

The open-source AI company’s latest world model generates high-fidelity interactive environments on hardware accessible to independent developers and researchers.

New research from IBM identifies systematic weaknesses in how AI agents reason and use tools, offering a framework for evaluating agentic system reliability.

Google’s open-weight Gemma 4 Vision-Language-Action model has been demonstrated running directly on NVIDIA’s Jetson Orin Nano Super, marking a step forward for on-device robotics AI.

Chinese AI lab DeepSeek has unveiled its latest foundation model featuring a one-million token context window designed for agentic applications.

Google launched Gemini 3.1 Flash Live, a model update designed to make real-time audio interactions more natural and reliable.

Google has launched new inference modes for its Gemini API, giving developers granular control over cost and reliability trade-offs.

Google has released Gemini 3.1 Flash TTS, a text-to-speech model it describes as the next generation of expressive AI speech synthesis.

Google announced its eighth-generation tensor processing unit lineup featuring two chips purpose-built for the demands of agentic AI systems.

OpenAI has rolled out a major update to its Codex application for macOS and Windows, adding computer use capabilities, in-app browsing, image generation, memory and plugin support.

OpenAI announced new enterprise partnerships with Accenture, PwC and Infosys to scale its Codex AI coding platform, which now serves 4 million weekly active users.

OpenAI has released a major upgrade to its image generation capabilities, featuring improved text rendering, multilingual support, and enhanced visual reasoning.

OpenAI introduced WebSocket connections and connection-scoped caching to its Responses API, reducing latency for developers building AI agent workflows.

OpenAI released the system card for GPT-5.5, detailing the company’s safety evaluations and risk mitigations ahead of the model’s broader availability.

OpenAI announced it will offer ChatGPT for Clinicians at no cost to verified U.S. physicians, nurse practitioners, and pharmacists.

OpenAI announced GPT-5.5, its most advanced AI model to date, with improved speed and capabilities for coding, research, and data analysis.

Welcome to WordPress. This is your first post. Edit or delete it, then start writing!