Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more
THE SO WHAT
Gemini 3.8 Flash "working harder" via more reasoning steps and iterative tool calls — at potentially higher cost — is Google making the trade explicit: better answers for more tokens. If you’re cost-sensitive, you’ll need per-workflow benchmarks to decide where 3.8’s extra depth actually pays back versus sticking with 3.7.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIThe AI skills gap is growing while schools wait for perfect answers
Education’s hesitation is widening the skills gap—graduates are arriving without hands-on AI experience while employers are already restructuring workflows around these tools. If you run a team, treat AI literacy as an in-house training problem, not something the education system will solve on your hiring timeline.
Applied AIJustice Department sides with OpenAI in publishers’ copyright fight
The DOJ framing AI training as fair use tied to national security is a strong signal that US policy may favor broad training rights, even as the EU AI Act pushes toward stricter copyright controls. Multinational teams should assume a split regime—US-friendly training, EU-tight usage and disclosure—and architect data, training, and deployment pipelines to respect that divide.
Applied AIWhile AI Industry Frets Over Safeguards, One Company Is Building a Model That ‘Doesn’t Say No’
A "doesn't say no" model is effectively a stress test of the emerging safety and liability norms—if it gains traction, regulators will feel pressure to codify boundaries rather than rely on voluntary guardrails. If you’re deploying AI in production, assume your risk baseline is set by the strictest regime you operate in, not by the most permissive model on the market.
Applied AINvidia’s stock is climbing as investors get more confidence in an expanding base of AI customers
Dell’s results showing AI hardware demand beyond hyperscalers mean the “AI customer” is now every large enterprise with a data center budget. If you run infra or procurement, assume GPU scarcity and pricing power persist longer than your current planning horizon.