News reporting focused on AI and machine learning, covering the companies behind these technologies, their real-world applications, and the ethical concerns they raise. This includes areas like generative AI (large language models, text-to-image and video), speech tech, and predictive analytics.
An anonymous provider is offering free access to a frontier-class AI coding model named Ox Alpha on OpenRouter, with no company claiming responsibility. The model's performance has drawn attention, but its origin remains unidentified. Fingerprinting points to Z.ai, the Chinese AI developer formerly known as Zhipu AI, though the company has not confirmed.
The agents that suddenly started working around Christmas 2025 were not the product of a single breakthrough. They were the result of two curves finally crossing: model capability and agent harness maturity. The harness, the invisible scaffolding of tools, context, and guardrails that turns a raw model into an agent, is now being absorbed into the models themselves. And the next phase, according to agentic engineer Dan McAteer, will turn the harness into something entirely different: an interface for human attention.
OpenAI and HubSpot have acquired creator-led media outlets TBPN and Futurepedia, respectively, signaling a strategic shift from renting attention via sponsorships to owning audiences outright. The deals highlight the growing value of direct audience relationships and distribution as feeder systems for corporate goals, from AI communication to software sales.
Brendan Foody, co-founder of AI talent platform Mercor, took to X to accuse Sequoia of using a 'dual-pricing' structure where it invests in two tranches at different valuations, inflating the headline number. Sequoia partner Shaun Maguire pushed back, calling it a market reality rather than a scam. The practice has drawn attention to how venture capital firms and founders manipulate perceptions of startup success.
A new paper systematically evaluates three variants of projection sharing in transformer attention: shared key-value, shared query-key, and a single projection. The authors found that sharing key and value projections performs on par with standard QKV attention while reducing KV cache by 50% with only 3.1% perplexity degradation. Combining this with grouped-query or multi-query attention can cut cache by up to 96.9%, enabling practical on-device inference.