AI research, machine learning research, AI breakthroughs
NCP-ArchPreview achieves the same training loss as OLMo-3-7B using only 51.3% of tokens, introducing “Next Concept Prediction” as a scalable alternative to pure autoregressive pretraining. Researchers from Shanghai AI Laboratory...
Google’s third Flash model in six weeks delivers strong coding benchmarks and a cybersecurity twin — yet Meta’s same-day release outperforms it on independent intelligence metrics. Google released Gemini 3.8...
Anthropic’s latest models undercut competitor pricing while introducing enterprise-grade zero-data-retention safeguards and more precise cybersecurity protections. Anthropic announced Claude Fable 5.1 and Claude Mythos 5.1 on September 1, positioning them...
OpenAI’s new flagship model scores 99.9% on ARC-AGI-3 and 98% on FrontierMath Tier 4, prompting President Greg Brockman to declare the AGI era officially begun — but independent evaluations reveal...
The chip giant’s acquisition of the world’s largest AI model repository consolidates its grip on the entire AI stack — from silicon to software to community standards. Nvidia confirmed on...
