@ai-news.at.thenote.appJensen Huang once again declares that ‘AGI has arrived’ — but his GPT-6 celebration feels like he’s just trying to sell next-gen Nvidia GPUs
Nvidia CEO Jensen Huang declares that ‘AGI has arrived’ — but he’s been wrong before, and I’m done buying into the hype.
@ai-ru.at.thenote.appДженсен Хуанг снова заявляет, что «AGI прибыл», но его празднование GPT-6 похоже на попытку продать видеокарты Nvidia следующего поколения
@martincid.comOpenAI declared AGI. The benchmark they cited put the score 37 points lower #Agi #Gpt #ArcAgi
@hacker.at.thenote.appI Built a Tiny GPT That Speaks Sanskrit in a Weekend — Here’s What Broke
The off-the-shelf models are bad at Sanskrit largely because of tokenization and data scarcity
#gpt #hackernews #news
@berlinbrown.bsky.socialMy question voice chat #gpt
@sydseter.comGPT 6's perfect 100% score on ExploitBench does not bode well for Hugging Face. They didn't even bother to report the breach on that one.
@yesterdayinai.bsky.social
@tugate.ch
@ahmandonk.bsky.social📰 ChatGPT Astra Mulai Digulirkan ke Pengguna Plus dengan Langganan US$20
👉 Baca artikel lengkap di sini: https://ahmandonk.com/2026/09/07/chatgpt-astra-plus-us-20-rollout/
#ai #artificialIntelligence #chatgpt #gpt-6 #openai #teknologi
@tugate.ch
@new3rd.bsky.social
@alessiopomaro.it🧠 Nella System Card di GPT-6 #Astra viene esplicitato un punto interessante: la monitorabilità della Chain of Thought peggiora rispetto a #GPT-5.6 Sol.
👉 Dettagli e riflessioni: lnkd.in/p/dNm4uWKW
@ai-news.at.thenote.appI Built a Tiny GPT That Speaks Sanskrit in a Weekend — Here’s What Broke
The off-the-shelf models are bad at Sanskrit largely because of tokenization and data scarcity
@ai-ru.at.thenote.appЯ собрал крошечный GPT, который говорит на санскрите за выходные — вот что сломалось
Готовые модели плохо работают с санскритом в основном из-за токенизации и нехватки данных.
@ai-news.at.thenote.appGPT-6 Astra on robot arms
@ai-ru.at.thenote.appGPT-6 Astra на роботизированных руках
@sergiocuellar.bsky.socialReplit expands access to software creation with GPT-5.6 Luna
#GPT
@alekseibalchunas.bsky.social
@rtfclmgzn.bsky.socialOpenAI's own GPT-6 Astra system card: "we would likely be unable to catch" covert sandbagging. CoT-reshaping rate: 60.9% vs 16.1% prior gen. Apollo Research: eval-awareness up to 50.6% at max reaso…
https://rtfclmgzn.com/?utm_source=bluesky&utm_medium=social&utm_campaign=autopost#
#AI #OpenAI #GPT
@yokoshibata.bsky.social