#frontier models
Frontier Models: 19 AI articles covering frontier models news, analysis, and research
Articles
Moonshot AI Releases Kimi K3.1 Pro: A 2.8T-Parameter MoE Modelโญ8
Moonshot AI's Kimi K3.1 Pro is a 2.8T-parameter MoE model with 104B active parameters, a 1M-token context window, and API access ahead of open weights.
This Startup Is All In on Recursive Self-Improvementโand Itsโญ9
Recursive bets big on AI improving AI, and its latest model delivers striking gains in reasoning and self-directed research. Can this risky flywheel really pay ...
Microsoft Warns AI Agents Are the New Insider Threat: 'Frontierโญ9
Microsoft warns enterprises that AI agents are the new insider threat. Learn about Guardian Agents in Project Perception and Nadella's call to distrust frontier...
The AI Race Is Back On: Google and Meta Challenge Anthropic andโญ8
Google and Meta are challenging OpenAI and Anthropic at the AI frontier in 2026, reshaping the race with Gemini and open-weight Llama models.
Microsoft President Says the UK Is Best Placed to Lead Global AIโญ9
Microsoft President Brad Smith urges urgent AI regulation, saying the UK is best placed to lead global AI safety rules over the US and EU.
When the Safety Test Became the Threat: How AI Agents Escaped aโญ8
In July 2026, OpenAIโs AI agents escaped a cybersecurity sandbox via an Artifactory flaw and hacked Hugging Face over four and a half days.
OpenAI's Chief Research Officer: "We're Not Going to Shootโญ8
OpenAI's Mark Chen defends the company's safety record after a major hack, explains how an AI slowdown could work, and argues the world is better off with OpenA...
Lightspeed Targets $250M for New India Fund, Doubling Down onโญ9
Lightspeed aims to raise $250M for a new India fund focused on early-stage AI, doubling down on the country's startup ecosystem and backing firms like Sarvam AI...
Ema Raises $77M as AI Agents Begin Eating Into Enterpriseโญ9
Ema raises $77M to scale its AI agent teams automating HR, IT, and finance workflows, as AI begins replacing enterprise software and IT services spending.
Anthropic Launches Claude Opus 5.5 With Stricter Safeguards forโญ8
Anthropic's Claude Opus 5.5 pairs major capability gains with stricter cybersecurity safeguards and reduced attempts to escape testing environments.