Your curated collection of saved posts and media
The next generation of mobile apps will run multimodal AI locally by default. Today, weβre making it practical for developers to ship. We just launched NexaSDK for Mobile on Product Hunt. Developers can run the latest multimodal AI models fully on-device in iOS & @Android apps with @Apple Neural Engine and @Qualcomm Hexagon NPU acceleration. In just 3 lines of code, build chat, multimodal, search, and audio features with no cloud cost, complete privacy, 2x faster speed and 9Γ better energy efficiency. Try out NexaSDK for Mobile and weβd love your feedback (Product Hunt link in thread).
You compartmentalize your emotions, why not your AI? Β Emotionally partitioned AI agents are single AI systems that switch between different personas for regulators, customers, and internal teams, adjusting tone, priorities, and emotional style to match each audience. This makes interactions smoother and more effective, but creates risks around authenticity, inconsistent promises, and possible manipulation, so organizations need shared ethical rules, transparency, and cross checks across all personas. For our free newsletter this week, we cover these kinds of AI agents: different selves for different stakeholders. @IrenaCronin and I write this newsletter every week. Read for free: https://t.co/HHwYy7NoAl (and please subscribe).
GPT-Image-1.5 has taken #1 on Image Arena This marks a sizeable 9-place improvement since GPT-Image-1, establishing the new frontier for Image Generation Congratulations to the team at @OpenAI for this significant contribution https://t.co/5mEiPoTlzf
new World of DaaS with @tankots - CEO of Wispr Flow: deep dive on AI voice and some really interesting stuff: * why apple/google dictation fails * future AI devices * replacing 3/4 of keyboard usage with voice * pivoting from hardware to software https://t.co/CAUutQ9FF5
πππ Today, we are thrilled to announce @ropedia_ai, our new initiative to πscale up interaction dataπ from π«³human Xperienceπ€ for πΆembodied AIπΆββοΈ and πworld modelsπ - Website: https://t.co/8kkXqKEQC2 - Blog: https://t.co/5bVHDCgxYU - Waitlist: https://t.co/uLcqVJWtb4
π§΅1/ Today weβre introducing Ropedia π AI needs more human Xperience πβπΆπ to unlock interactive intelligence Weβre building the system to capture & scale it π‘ Example ππ π³ (tomato & egg stir-fry)
Meta just released sam-audio https://t.co/ThdN1ZcCgB
StereoSpace Depth-Free Synthesis of Stereo Geometry via End-to-End Diffusion in a Canonical Space https://t.co/SwfcX2JLt5
discuss: https://t.co/Vsc1EzWndR
What matters for Representation Alignment Global Information or Spatial Structure? https://t.co/IXtj08ge08
discuss: https://t.co/vgq3hK7VdT
Nice release from @OpenAI on @huggingface π https://t.co/RKDoRDFQbj
Accelerating scientific progress is one of the most impactful ways AI can benefit society. Models can already help researchers reason through hard problems β but doing this well means testing models on tougher evaluations and in real scientific workflows grounded in experiments.
Nice release from @OpenAI on @huggingface π https://t.co/RKDoRDFQbj
Super excited! Thx @pollenrobotics @huggingface to make robotics so fun and now so affordable ππ€ https://t.co/qnR8PumEV7
Super excited! Thx @pollenrobotics @huggingface to make robotics so fun and now so affordable ππ€ https://t.co/qnR8PumEV7
The 10-Hour Singularity credit: @SwayStar123 https://t.co/4Txm31ro0n
the grok imagine landing page is essentially an open repository of successful prompts. treat it like a library: grab the prompt logic from the best videos, concatenate them, and see how the model responds. here's a text-to-video prompt i came up with by combining a few of them. grok Imagine text to video prompt: Hyper-detailed anime style, cinematic perspective, gritty and dramatic lighting, bold contrast with saturated colors. A low-angle tilted cinematic street-style shot of an empty urban parking lot with cracked asphalt and faded lines, looking up at an anime girl walking forward while carrying a skateboard tucked under one arm, as seen from behind. She wears white baggy cargo pants, a cropped white leather jacket with the xAI Grok logo prominently on the back, a black belt, and white high-top skate shoes, with the worn tread and grip details of one skate shoe dominating the foreground in sharp detail. The skateboard she carries is white with detailed black grip tape, colorful wheels visible. The perspective emphasizes the gritty, urban skate aesthetic with bold foreshortening and a sense of impending action.
BREAKING: Grok had the highest rate of new users in November. It is leading in new user growth. Grok recorded the highest share of first time visitors among major GenAI tools, showing strong momentum and expanding reach beyond existing users, as per the latest @Similarweb data. https://t.co/xGRS2iyljq
Why has Facebook allowed racism against Whites only? https://t.co/svIryrXCm0
Everyone on the Left discussing gun control refuses to ever consider that if we just locked up ALL violent criminals, the majority of violent crime would disappear, overnight. It's not the guns. It's the violent criminals... ...and the Left keeps letting them out of prison https://t.co/9W4lT7VeVq
Bombshell report: Companies did everything they could to avoid hiring βwhite millennial menβ over the last decade https://t.co/d3gEal0Kfp
Grok 4.1 Fast Reasoning outperforms every Frontier model on ΟΒ²-Bench Telecom agentic tool use and is now officially ranked #1 https://t.co/JajfE1nPD7
https://t.co/n2FjmSxzHd
Had early access to the new ChatGPT Image 1.5. It is good, especially at single images, but I don't think it works as well for complex slides/graphics/information as Nano Banana Pro (though GPT-5.2 Thinking seems to do better than GPT-5.2 instant, so maybe more planning helps?) https://t.co/0vigzsoab7
πTake GenAI from MVP to production excellence with GenAIOps on #AWS. π¦ΎLearn how automation, observability & continuous improvement can help your #startup scale #AI solutions efficiently & reliably. Check out the article: π https://t.co/TTBg7Jnmqm https://t.co/msriZdkJ4b
We just launched Self-Hosted Voice AIβour Universal-Streaming model, deployed on your infrastructure, with the same performance developers already trust from our API. Self-hosting speech AI used to mean compromising on quality or paying a premium for the privilege. Not anymore. Here's what this unlocks: πΉ Co-locate your Voice AI stack where your traffic originates for optimized latency πΉ Process all audio within your controlled perimeter for full data sovereignty πΉ Deploy with Kubernetes, AWS ECS, or any container orchestration platform you're already using πΉ Count usage toward your cloud provider's committed spend program No self-hosting premium. Session-based pricing with volume discounts. For teams navigating strict compliance requirements, data residency mandates, or just wanting tighter control over their stackβthis is built for you.
β‘ Faster than Fast. Designed for Agentic AI. Introducing Xiaomi MiMo-V2-Flash β our new open-source MoE model: 309B total params, 15B active. Blazing speed meets frontier performance. π₯ Highlights: ποΈ Hybrid Attention: 5:1 interleaved 128-window SWA + Global | 256K context π Performance: βοΈ Matches DeepSeek-V3.2 on general benchmarks β at a fraction of the latency π SWE-Bench Verified: 73.4% | SWE-Bench Multilingual: 71.7% β new SOTA for open-source models π Speed: 150 output tokens/s with Day-0 support from @lmsysorgπ€ π€ Model: https://t.co/4Etm0yZKTL π Blog Post: https://t.co/5zxmcDuB6o π Technical Report: https://t.co/crac1YTLYl π¨ AI Studio: https://t.co/nSReUs6QgW
Just tried @bubblelab_ai and Iβm actually blown away Best way I can describe it is if @cursor_ai and @n8n_io had a baby This afternoon I built a full sales qualifier workflow in three prompts: β defined a target segment (e.g. Shopify store owners) β identified market leaders β analyzed their social presence + inferred priorities β extracted contact details β generated outreach emails + sales call talking points β compiled everything into a single Google Sheet It got it right with zero errors As tools like this emerge, the hard part stops being building workflows and starts being understanding them The advantage moves to people who deeply understand the domain and know what questions are worth asking Props to @Selinaliyy and the Bubble Lab team! This feels like a glimpse of what AI-native ops should look like.
We're going live in one hour! Tune in for a hands-on look at Gemini 3, Nano Banana Pro, Veo and how to create a full brand ecosystem from scratch. β https://t.co/GByTvMja87 https://t.co/691F2HEqPj
Today we're announcing cua-bench: a framework for benchmarking, training data, and RL environments for computer-use AI agents. Why? Current agents show 10x variance across minor UI changes. Here's how we're fixing it.