OpenAI
OpenAI shows the chip that promises much faster AI responses
August 25, 2026 · OpenAI, The Verge and TechCrunch
More work per watt and lower latency for agents and models, with initial deployment still limited.
Filtered news
The section does not want to be an automatic feed. Each piece must explain what changes, who is affected and what is worth trying next.
Enter: Changes in price, limits, privacy, availability in Spain, tools that can already be used, and measurable improvements in real tasks.
Don't enter: Rumors, unannounced news, repeated press notes and headlines that don't change any practical action.
Method: primary sources before headlinesSufficient context and a practical conclusion separate the news.
OpenAI
August 25, 2026 · OpenAI, The Verge and TechCrunch
More work per watt and lower latency for agents and models, with initial deployment still limited.
Claude
08/25/2026 · TechCrunch, VentureBeat and Anthropic
Shared memory between chat and Cowork reduces context repetition, but requires looking at privacy.
Gemini
August 25, 2026 · Google Blog
Natural speech dictation in macOS to speak naturally, clean phrases and put the result directly.
Google News
August 20, 2026 · Google Blog and The Verge
Language preference settings, favorite font sources and news audio configurations.
Research
August 24, 2026 · MIT News and Nature Communications
Scenarios that generate plausible extreme scenarios for infrastructure, climate and markets.