I think most people who are not playing with these technologies are way underestimating the speed of advancement in "creative" AIs & how much of an impact that is going to have on many jobs, soon. AI language & art models are growing at 10x a year. Beyond Moore's Law pacing. 1/ https://t.co/r0YnROh58z

everyone is writing code but no one is making money https://t.co/PTQh6kZDCT

you asked for it, you get it @SherryYanJiang and i are finally bringing cafe @cursor_ai singapore aka cursor kopitiam to town @nickwm and @fr4nnyp4ck will be flying in to drink some coffee and hang out spend the evening hacking or just catching up with other cursor-pilled folks registration linked below:

@abrakjamson https://t.co/Def9FH8dh1

Very proud of this amicus brief filed yesterday in the @AnthropicAI case against the Department of War from Catholic moral theologians and ethicists. The very notion of what it means to have a just war is at stake in how we respond to these matters. https://t.co/Fpm0Ercx40 https://t.co/NvbIoJA8xS

NVIDIA has released Nemotron 3 VoiceChat! A ~12B parameter Speech to Speech model that leads our open weights Conversational Dynamics vs. Speech Reasoning pareto frontier Understanding Speech to Speech model performance is multidimensional - two key and distinct dimensions are raw intelligence and conversational dynamics: how well a model handles the natural rhythms of human conversation such as turn-taking, interruptions. Amongst full duplex open weights models, NVIDIA’s new Nemotron 3 VoiceChat, V1, leads in balancing these dimensions, setting itself apart from other models on the Conversational Dynamics vs. Speech Reasoning pareto frontier. Key benchmarking results: ➤ Conversational Dynamics (Full Duplex Bench): Nemotron 3 VoiceChat (V1) scores 77.8%, second among open weights speech to speech models behind NVIDIA's own PersonaPlex (91.0%) and ahead of FLM-Audio (62.0%), Moshi (61.0%) and Freeze-Omni (58.7%) ➤ Speech Reasoning (Big Bench Audio): Nemotron 3 VoiceChat (V1) scores 29.2%, second among open weights speech to speech models behind Freeze-Omni (33.9%) and well ahead of PersonaPlex (12.6%), FLM-Audio (5.3%) and Moshi (1.7%) ➤ Pareto leader: While Freeze-Omni leads on speech reasoning and PersonaPlex leads on conversational dynamics, Nemotron 3 VoiceChat (V1) is the only open weights model that performs amongst the top 3 on both - making it the clear leader on the pareto frontier between these two critical dimensions ➤ Larger than other open weights models but still relatively small compared to LLMs: Nemotron 3 VoiceChat (V1) has 12B parameters, making it one of the larger open weights speech to speech models, while NVIDIA's PersonaPlex is ~7B. While larger compared to other larger open weights speech to speech models the model still is relatively small compared to leading LLMs ➤ Context vs. proprietary models: While this release materially advances open weights performance, open weights speech to speech models still significantly underperform leading proprietary offerings. For comparison, proprietary models on our Big Bench Audio benchmark score substantially higher - Step-Audio R1.1 at 96%, Grok Voice Agent at 92%, Gemini 2.5 Flash (Thinking) at 92%, and Nova 2.0 Sonic at 87%. The gap between open weights and proprietary remains large in this modality. As the capability and adoption of Speech to Speech models increases, we expect to expand our set of benchmarks to include elements such as tool-calling and multi-turn instruction following. See more details below ⬇️

Jensen today announced Alpamayo 1.5 at #NVIDIAGTC! #Alpamayo 1.5 is a major update to Alpamayo 1—@nvidia’s open 10B-parameter chain-of-thought reasoning VLA model, first introduced at #CES. Built on the #Cosmos-Reason2 VLM backbone and post-trained with RL, it adds support for navigation guidance, flexible multi-camera setups, configurable camera parameters, and user question answering. The result is an interactive, steerable reasoning engine for the AV community. We’re also releasing post-training scripts to help researchers and developers adapt the model. Additionally, we’ve significantly expanded the Alpamayo open platform across data and simulation, including releasing highly requested reasoning labels for the PhysicalAI Autonomous Vehicles dataset (https://t.co/fD9eUcndya), as well as our chain-of-causation auto-labeling pipeline. 🔎 Learn more about Alpamayo 1.5 and the latest extensions to the Alpamayo open platform: https://t.co/P0nuqkwBab (please note that most of the links will become active in the next few days.) Happy building—and stay tuned for more in the coming months! @NVIDIADRIVE @NVIDIAAI

+1 more

NVIDIA officially announced its expansion into space computing, introducing a suite of accelerated platforms designed to power orbital data centers (ODCs) and autonomous spacecraft. The lineup features the new Space-1 Vera Rubin Module, which delivers 25x more AI compute for space-based inferencing compared to previous models, alongside the energy-efficient IGX Thor and Jetson Orin edge computing modules. Engineered to survive the strict size, weight, and power (SWaP) constraints of space, these platforms allow industry partners like Axiom Space, Planet Labs, and Starcloud to process massive amounts of geospatial and sensor data directly in orbit, rather than beaming raw data back to Earth.

NEWS: SoundHound AI Unveils World’s First Multimodal Agentic+ AI Completely on the Edge at NVIDIA GTC 2026 🤖 🔗 https://t.co/3jGB3JKZMC https://t.co/iaEewsBVcz

🛠️ Claude Code "opusplan" 말 그대로 하이브리드 모델.. 공식임! Claude Code에는 opusplan 모델을 선택할 수 있어요. > /model opusplan 하이브리드 모델 alias인데, 작업 단계에 따라 자동으로 모델을 전환해요. 복잡한 추론을 위한 플랜 모드에서는 Opus를 실행 단계에서는 Sonnet으로 자동 전환됩니다! Opus로 계획하고 구현까지 하는 것도 물론 가능해요. 하지만 이미 탄탄한 계획이 있다면, 실행은 Sonnet으로도 충분하고 더 저렴할 수 있어요. 각 작업에 맞는 모델을 쓰는 것 = 효율 🚀 플래닝과 실행은 요구되는 인지 부하가 달라요. Opus의 깊은 추론 능력은 계획 수립 단계에서 가장 빛나고, 일단 탄탄한 계획이 세워진 이후의 실행은 Sonnet으로 충분히 커버될 수 있어요. 언제 쓰면 좋냐구요? - 복잡한 기능 설계같이 아키텍처 결정이 중요한 작업 - 리팩토링 계획같은 영향 범위 분석이 필요한 경우 - Opus 풀파워 사용 대비 비용 절감이 필요할 때 이거 이제 많이 활용하실 듯!!!

NVIDIA Vera Rubin achieves 40 million times more compute in 10 years: 288 GB HBM4, 22 TB/s bandwidth, 50 PFLOPs of AI horsepower. https://t.co/Nkiu06RaB9

AI really can help education: Randomized controlled experiment on high school students found a GPT-4o powered tutor that personalized problems for students raised final test scores by .15 SD, "equivalent to as much as six to nine months of additional schooling by some estimates" https://t.co/QAUHCXRzAn

Sakana AIの技術を武器に、顧客の現場に飛び込んで直接ビジネス課題を解決する「Forward Deployed Engineer (FDE)」を募集します🐟 https://t.co/I4LQlKPr5z 単なるAI導入ではなく、RAGやAIエージェントなどのコンテクストエンジニアリングを駆使し、顧客の泥臭い課題をAIで解ける形に構造化する ポジションです。Applied Teamの最前線で、顧客の頼れる相談役となり、AIの社会実装を加速させてくれる方を探しています！ 技術力とビジネス解決力の両方に自信がある方、ぜひご応募ください。

Sakana AIの技術を武器に、顧客の現場に飛び込んで直接ビジネス課題を解決する「Forward Deployed Engineer (FDE)」を募集します🐟 https://t.co/I4LQlKPr5z 単なるAI導入ではなく、RAGやAIエージェントなどのコンテクストエンジニアリングを駆使し、顧客の泥臭い課題をAIで解ける形に構造化する ポジションです。Applied Teamの最前線で、顧客の頼れる相談役となり、AIの社会実装を加速させてくれる方を探しています！ 技術力とビジネス解決力の両方に自信がある方、ぜひご応募ください。

Look who we spotted. Huge moment at #NVIDIAGTC as @MichaelDell and Jensen Huang stopped by our booth where Jensen signed the next generation, 100% liquid cooled PowerEdge XE9812 Vera Rubin NVL72 server. https://t.co/3423zH0pSx

https://t.co/dxjyCCX9DQ

https://t.co/dxjyCCX9DQ

https://t.co/EEWgdDipOE

https://t.co/EEWgdDipOE

https://t.co/33TjpZvDMl

@JasonBotterill https://t.co/C4FrLe3gxZ

Bing as a universal folklore machine It can look up folktale styles using the Aarne-Thompson-Uther classification of world folktales (though it sometimes gets the exact index number wrong), and then rewrite them for imaginary cultures: from post-apocalyptic nomads to sky pirates https://t.co/Ncb2owzqEc

Gemini as folklore machine: "Create a comic using universal folklore index ATU 570 set in the present day" "Now add ATU 720" ATU 570 are tales about "the king's rabbit herder" & ATU 720 is "My mother slew me, my father ate me, my sister buried me under the juniper" (really) https://t.co/tXkmDYLc1Y

I suspect I am getting a bit obscure in my tests at this point, but it is cool that the AI can look up folklore numbers from this massive index of world folklore motifs and turn them into comics, making the weird incoherent stuff built into folklore make some sort of sense. https://t.co/ozYyGXgpNB

Visiting the NVIDIA GTC expo hall shows you why OpenAI is changing strategy. All the big booths are enterprise. Does this mean @sama is gonna give up his dreams of stealing Tim Cook’s lunch? The biggest news here is how NVIDIA is bringing @openclaw to the enterprise. The hidden message is “you don’t want to run your business on a Mac Mini, do you?” Or, even more hidden: “a stack of DGX computers will be better for your development team since you can run an NVIDIA model on them and cut your token-generation bills with Anthropic down.” But if I can discern these hidden messages at GTC so can others. One guy sitting next to me remembers when NVIDIA was for gamers. Barely a mention of video games here this year. But the Holodeck team has a McLaren right in the middle of the NVIDIA booth. With a cool Apple Vision Pro app to do air flow analysis on a race car. Here is Max from NVIDIA telling me about it. I have been telling everyone that Apple has the Formula 1 exclusive, which will keep Apple relevant even if OpenAI went consumer. Enterprise is the smart move. That video game stuff was last week at GDC. I am doing a ton of video this week for Chinese media and investors: https://t.co/2BjIVhqvR4 OpenClaw is even hotter in China than in USA. Why? An analyst I met explained it this way: the Chinese heard that Anthropic hates them, so are choosing systems that let them use their own models. Global politics are everywhere on the show floor. Oh, and same guy said he knows developers who are spending $1,000 a day with their OpenClaws. That is certainly enterprise. Consumers look at you like you have a mental illness if you try to tell them they need to spend $1,000 a day on tokens. But someone is, which is why Jensen said he raised his revenue projections to a trillion dollars. Ask your enterprise developers are throwing cash into the AI factories to get those tokens out. What an industry. Enterprise is sexy all of a sudden. Didn’t see that coming.

Closeted gay men before they have sex with a woman for the first time https://t.co/QhqqWRneFQ

Closeted gay men before they have sex with a woman for the first time https://t.co/QhqqWRneFQ

Packed house for the @Modular x @llama_index private dinner listening to @jerryjliu0 + @clattner_llvm share their thoughts on everything happening in AI + here at GTC https://t.co/9W1KPdRPqY

NYC spends more per homeless person than the median NYC household earns. $81,705 per person in FY2025. And $81,705 is a floor. It excludes supportive housing (~$500M/yr), mental health response teams, and NYPD encampment costs. The city projects ~$97K per person in FY2026. https://t.co/haXJenA4Q3

There's an engineer on YouTube building his own room-scale laundry-picking UFO catcher robot out of QR codes and string, it's one of the most compelling robotics demos I've seen in a while. https://t.co/QuASlOi4Eh

The moment he realised that https://t.co/vWmBsnR1nt isn't fully built on transformers and we can run on a single GPU with high accuracy and lower cost https://t.co/ZJYuL62UB8

The moment he realised that https://t.co/vWmBsnR1nt isn't fully built on transformers and we can run on a single GPU with high accuracy and lower cost https://t.co/ZJYuL62UB8