Now, Tencent claims its Hy4 Preview has reportedly outperformed the models of these newer AI upstarts. Primarily aimed toward coding, game development, and scientific research, the model works on 770 billion parameters, has 49 billion active parameters per token, and a token context window exceeding 1 million.

The AI company Runway has unveiled a new type of AI model it’s calling an interface world model, allowing you to create interactive websites and more by directly interfacing with elements on screen.

Meta says it has caught up with Anthropic and OpenAI after releasing Muse Spark 1.3, its most powerful LLM so far - SiliconANGLE

Available to developers via the Meta Model API at the same price as its predecessor, the paid model will soon integrate into Meta's flagship consumer products, including Instagram, Facebook, and Meta AI.

AI success is no longer about the biggest model. Winning companies will match frontier, open, and specialized AI models to each task.

Meta has introduced its first real-time audio model, Muse Voice Transcribe. The model can handle dictation and transcription for more than 20 speakers and can seamlessly handle multiple languages at once, Meta says.

Z.ai’s latest AI model release could help companies secure their systems—or find its way into the hands of hackers.

VMware’s VCF leader explains frontier AI model security and innovation, new software patches strategy, and VMware Explore 2026.

Naver Cloud has been finally selected as the business operator to develop an AI (artificial intelligence) foundation model specialized in cybersecurity. This project was initiated to address growing security threats as AI becomes more sophisticated and is increasingly exploited in cyberattacks,

At the foundation of the new capabilities is Eluvio's Universal & Dynamic Video Intelligence Architecture, which integrates AI inference, multimodal intelligence, text and vector storage, search, generation, and agentic orchestration directly into the Content Fabric media pipeline. Its key capabilities include:

Researchers on the company’s alignment team, the group whose job is to check that its models behave as intended, named it Hacker-Opus.