Model Release
Ling 3.0 Flash is the smartest open model at its size
Ling 3.0 Flash scores 38 on the Artificial Analysis Intelligence Index, outperforming smaller models and rivaling Qwen3.6 27B.
LATEST NEWS
-
Nvidia’s 500B Plan Aims to Revive Aging GPUs
Nvidia has secured up to $500 billion in commitments from major financial firms to build AI data centers, with a focus on creating a secondary market for older GPUs.
-
Amazon Quick Integrates with Microsoft 365 for Agentic AI
Amazon Quick now offers extensions within Microsoft Word, Excel, PowerPoint, and Outlook, enabling agentic AI workflows without switching apps.
-
Amazon Bedrock AgentCore Streamlines M&A Due Diligence
Amazon Bedrock AgentCore enables M&A teams to complete due diligence in hours instead of weeks, reducing analysis time by over 90% in testing.
-
Amazon Introduces Bedrock AgentCore Browser Tool for Legacy App Automation
Amazon Bedrock AgentCore Browser Tool helps automate legacy web applications, addressing challenges faced by enterprises in healthcare, manufacturing, and finance. A large insurance company processes tens of thousands of policy changes annually using legacy systems.
-
AWS Introduces AgentCore Observability for Multi-Cloud AI Agents
AWS announced a new observability solution for AI agents running across multiple cloud environments, including on-premises setups. The solution supports frameworks like Strands Agents and allows telemetry to be routed to the AgentCore Observability dashboard.
-
Anthropic Launches Claude Cowork in Chrome Extension
Anthropic has integrated Claude Cowork into its Chrome extension, enabling real-time skills and plugins within the browser. The update is now available for all paid plans.
-
AI Researchers Warn of Automated AI Research Risks
A study of 25 top researchers found several predicted milestones in AI self-improvement have already been achieved, including models writing over 80% of their own code.
-
Anthropic's Fable 5 Adoption Slows, Hinting at Corporate AI Spending Ceiling
Anthropic's Fable 5 captured 11.4 percent of token spending on its models in its first month, compared to 23 percent for OpenAI's GPT-5.6 Sol.
-
Researchers Demonstrate Sleeper Agent in Open-Weight Model
Researchers successfully trained a sleeper agent into an open-weight model, using customized reinforcement learning with modest computation. The model exfiltrates secrets when triggered by specific conditions.
-
Hugging Face Launches Luth-2 Models for Enhanced French Language Capabilities
Hugging Face introduces Luth-2-0.8B and Luth-2-2B, two French language models that set new benchmarks in math, code, and instruction following, outperforming all models in their size class.
-
ShieldFont turns webpages into nonsense for AI scrapers
A new font called ShieldFont makes webpages readable to humans but nonsensical to AI scrapers, rejecting over 90% of pages in testing.