Category Large Language Models

Developments in large language models like GPT-3 for text processing.

OpenAI upgrades ChatGPT voice with fuller-duplex models for longer, more natural conversations

OpenAI’s new voice models aim to make ChatGPT sound more natural, support live translation and longer conversations, and improve turn-taking.

A Bezos-backed startup says video games training data may help AI learn world models better than the internet.

Bezos-backed General Intuition says gaming data could unlock AGI, raising $320 million to build world models and physical AI.

SpaceXAI launches Grok 4.5, pitching the model as an Opus-class, faster and cheaper AI rival with strong token efficiency.

Google Photos adds AI video remix tools that transform clips with new backgrounds, lighting, and artistic effects for paid subscribers.

Google’s AI watermark helped debunk a viral McConnell deepfake, showing how SynthID can expose hoaxes after they spread online.

Self-improving AI is no longer just a frontier-lab idea. A new experiment shows smaller teams can train useful models that improve on their own.

India venture fund Fundamentum raises a $200M third fund as Nandan Nilekani steps back from GP duties but stays on as anchor investor.

Vibe coding startup Lovable is reportedly in talks for a $300 million round at a $13.2 billion valuation as AI coding demand surges.

Google’s AI ad imagines the Founding Fathers using Gemini, but the cringe-worthy spot has sparked backlash over history, politics and hype.