the benchmarks aren't close. they're embarrassing. > humanity's last exam: kimi 44.9%, beats closed models
> browsecomp (agentic web search): kimi 60.2% vs gpt-5's 54.9%
> gpqa diamond: 85.7% vs gpt-5's 84.5%
> swe-bench verified: 71.3% (coding tasks)
> artificial analysis
TECHNOLOGY
-
Kimi outperforms GPT-5 on key benchmarks like Humanity’s Last Exam
By
–
-
China’s $4.6M model beats GPT-5; Moonshot’s Kimi K2 triumphs
By
–
china just dropped a $4.6M model that beats gpt-5. while silicon valley burned billions. moonshot's kimi k2 thinking: – 1 trillion parameters trained for less than a nice house in palo alto. – beats openai's flagship on humanity's last exam (44.9% vs proprietary models)
– -
AI builds sales pitch, researches prospect, emails deck in 90 seconds
By
–
I just watched an AI build a complete sales pitch, research the prospect, and email them a personalized deck.
— God of Prompt (@godofprompt) 10 novembre 2025
All in 90 seconds and its 100% automated.
Sales teams are about to get absolutely decimated.
Here's how to do it + comment "AI" and I'll DM you my automation guide: pic.twitter.com/WnkiWx3H8vI just watched an AI build a complete sales pitch, research the prospect, and email them a personalized deck. All in 90 seconds and its 100% automated. Sales teams are about to get absolutely decimated. Here's how to do it + comment "AI" and I'll DM you my automation guide:
-
EU commits €1.4 billion for deep tech innovation in 2026
By
–
Commission adopts 2026 European Innovation Council work programme, making €1.4 billion available for deep tech innovation | Shaping Europe’s digital future https://
digital-strategy.ec.europa.eu/en/news/commis
sion-adopts-2026-european-innovation-council-work-programme-making-eu14-billion-available
…
#digitaleu #innovation #technology @DigitalEU @elaniazito @ArturHabant @CurieuxExplorer @Shi4Tech -
CEO Redoes Robot Demo Amid Doubts Over Human in Suit
By
–
Le nombre de commentaires qui parlent d’unijambiste est hallucinant. Le PDG n’en dormait plus : il a refait une démo, cette fois en découpant les deux mains et en déchirant le costume de partout.😅
— VISION IA (@vision_ia) 10 novembre 2025
Certains diront que c'est un humain Amputé quadruple dans une combinaison… https://t.co/ktUOUN9G88 pic.twitter.com/tPkgk5iAeRThe number of comments talking about a one-legged person is mind-blowing. The CEO couldn't sleep anymore over it: he redid a demo, this time by cutting off both hands and tearing the costume everywhere. Some will say it's a quadruple amputee human in a robotic suit. Logical,
-
Future of Medical Imaging: Daily Full Body MRI Scanning
By
–
I can’t wait for imaging to get better. My full body MRI was expensive and took an hour strapped down. I can’t imagine doing it more than once every few years. When will we have something we can just scan with every morning?
-
Google’s new model blurs the line between real and not
By
–
This new model definitely blurs the line further of what’s real or not. Incredible work by Google.
-
Full Body MRI Results: AI Healthcare Imaging Application
By
–
Got my full body MRI results. So cool. pic.twitter.com/aQ2HhrOay1
— Aaron Ng (@localghost) 9 novembre 2025Got my full body MRI results. So cool.
-
AI job apocalypse is just a negotiation tactic
By
–
every CEO: “AI will replace 30% of jobs” *implements AI* *productivity drops*
*costs go up*
*rehires humans* MIT study: 95% of AI pilots failed. Fed data: only 1% actually did layoffs. the AI job apocalypse is a negotiation tactic wrapped in a press release. they want you -
Rolling Forcing Enables Sharp, Coherent Long-Horizon Video Generation
By
–
AI can stream videos that stay sharp and coherent for minutes.
— 机器之心 JIQIZHIXIN (@jiqizhixin) 9 novembre 2025
Meet Rolling Forcing—a new technique for long-horizon video generation that slashes error accumulation.
It denoises multiple frames jointly, anchors long-term context via an attention sink, and trains efficiently… pic.twitter.com/E7QitWyGrDAI can stream videos that stay sharp and coherent for minutes. Meet Rolling Forcing—a new technique for long-horizon video generation that slashes error accumulation. It denoises multiple frames jointly, anchors long-term context via an attention sink, and trains efficiently