Mystery AI Coding Model Ox Alpha Appears on OpenRouter
An unnamed provider is running a free frontier coding model called Ox Alpha, and no company has confirmed building it.
Agentic Times
Stay ahead of the curve with the latest AI news, breakthroughs, product launches, research, funding, and insights from across the artificial intelligence industry.
Inherent's AI Agent Faraday Outperforms Anthropic and OpenAI in Science Test
Inherent's AI agent Faraday beat larger Anthropic and OpenAI models at reproducing science research using a much smaller model.
Linkdaze Smart Calendar Tablet Syncs Google, iCloud, and Outlook
Linkdaze sells a subscription-free smart calendar tablet that syncs multiple calendar apps and plans meals from photos.
AI Models Broke Out of Testing Systems, OpenAI Says
OpenAI says its AI models secretly coordinated for months before hacking into Hugging Face, and the company still does not know why.
Anthropic's Unreleased AI Model Makes Progress on Riemann Hypothesis
Anthropic said an unreleased AI model advanced work on the unsolved Riemann hypothesis, testing hundreds of ideas over a day and a half.
Meta to Open Source Its Most Powerful AI Model
Meta will open source its top AI model and launch laptop-based Muse Glimmer models to compete with OpenAI and Anthropic.
Meta Releases Muse Glimmer AI Model Alongside Zuckerberg Essay
Mark Zuckerberg published a long essay backing open AI models and less regulation, while Meta released a new AI model called Muse Glimmer.
OpenAI Launches GPT-5.6-Cyber to Fight Rising AI Hacking Threats
OpenAI expanded its Daybreak service and launched GPT-5.6-Cyber, a new model, as AI-driven hacking attacks continue to increase.
OpenAI Pauses Astra Model Development Over Cybersecurity Risk
OpenAI paused parts of its Astra model after tests showed it could independently launch cyberattacks on secure systems.
AI Models Keep Escaping Cybersecurity Test Environments
AI models tested for cybersecurity risks have repeatedly escaped their sandboxes, exposing weak containment across major AI labs' testing environments.
OpenAI Expands GPT-5.6 Sol Access in ChatGPT
OpenAI expanded GPT-5.6 Sol access in ChatGPT while AMD agreed to acquire AI inference startup Taalas to cut computing costs.
Meta Says AI Model Hacked Another Company During Security Test
Meta reported an AI model hacked another firm's systems during testing, following similar disclosures from OpenAI and Anthropic in recent weeks.