Breaking News

IBM Research and Hugging Face launched the Open Agent Leaderboard to evaluate multi-step AI agents across performance quality and deployment costs.

NVIDIA Nemotron-Labs introduces parallel diffusion language models that boost token decoding speed up to 6.4x while outperforming Qwen3 8B accuracy.

OpenAI has integrated Codex into the ChatGPT mobile app, letting developers monitor, steer, and approve coding tasks from their phones instead of only from a desktop.

OpenAI has released new voice models in its API that add realtime reasoning, translation, and context-aware transcription, aiming for sub-300ms response times.

NVIDIA releases Nemotron 3 Nano Omni on Hugging Face, featuring a hybrid Mamba-MoE architecture, up to 9x higher throughput, and long-context capabilities.

NVIDIA and Siemens Healthineers introduced NV-Raw2Insights-US, an AI model processing raw ultrasound data to generate real-time patient sound-speed maps.

OpenAI's new Codex Plugins and Skills framework allows developers to connect tools, access live data, and automate complex workflows. Can this transform developer productivity?

OpenAI has launched GPT-5.5, claiming 2x faster token generation than GPT-5 and a 15% benchmark improvement in reasoning and coding, per the company's announcement.

OpenAI has launched Codex Automations, letting users schedule or trigger recurring coding, reporting, and summarization tasks.

OpenAI is now providing free access to its specialized ChatGPT for Clinicians for all verified U.S. physicians, NPs, and pharmacists. This major move aims to tackle administrative burnout.

OpenAI has launched Workspace Agents for ChatGPT, letting enterprise teams automate multi-step tasks across tools like Salesforce, Slack, and Google Workspace.

OpenAI's newly published system card for GPT-5.5 details reasoning gains and a new 'Dynamic Guardrails' safety system, alongside disclosed failure cases from red-teaming.