What really changes the game: → Nano Banana Pro can render sharp, accurate text in images (logos, UIs, mockups, ads). → Gamma’s agent lets you make bulk edits across slides, pages, and docs instead of fixing one by one. https://t.co/JTAwQO3xDv

And for those who want to level up with Gamma: Nano Banana Pro HD in Gamma’s Ultra plan gives you 4K‑level quality for decks and sites. https://t.co/rNCSWT5cVb

Oh, and the best part? Nano Banana Pro is now free to use, regardless of your tier. Get started here: https://t.co/MfeVcxTvof https://t.co/Qbv0NjKQ9d

Something big is happening in robotics - and it’s hiding in plain sight. This post is not about dancing robots but in the data that powers them. Open robotics datasets have exploded this year, turning the field into a more scalable and collaborative ecosystem. In just two years, @huggingface datasets grew from 11k to over 600k - and robotics is by far the fastest-growing segment. We went from 1k robotics datasets in 2024 to 27k in 2025! For comparison, text generation, the second-largest category, has only around 5k datasets in 2025. That gap is massive. Open datasets are important because robotics lives and dies by real-world robot data - video, actions, sensors, failures. By making this data easy to upload, reuse, and benchmark, researchers, startups, and large players are now releasing real-robot datasets that would have stayed locked inside labs just a few years ago. Major contributors include @nvidia, LeRobot initiative, and a rapidly growing maker community. This surge is also enabled by cheaper video storage, better tooling, and an open-source AI culture now spilling into the physical world. And it really matters: open robotics data dramatically lowers entry barriers, accelerates learning-by-doing, and speeds up progress toward generalist and humanoid robots. Robotics won’t scale through hardware alone - but to a large extent through shared data. Viz below from @aiworld_eu - link to the story and more viz/filters in comment.

🚀 New on ModelScope: QwenLong-L1.5 is now fully open-source! A 30B model (3B active params) that matches GPT-5 & Gemini-2.5-Pro in long-context reasoning. 🔥 Key wins: ✅ +31.7 pts on OpenAI’s MRCR (128K context → SOTA across all models) ✅ Matches Gemini-2.5-Pro on 6 major long-QA benchmarks ✅ +9.69 on CorpusQA, +6.16 on LongBench-V2 How? Three breakthroughs: 1️⃣ Synthetic data at scale: 14.1K long-reasoning samples from 9.2B tokens — no human labeling. Avg. length: 34K tokens (max: 119K!). 2️⃣ Stable RL training: Task-balanced sampling + Adaptive Entropy-Controlled Policy Optimization (AEPO) for reliable long-sequence learning. 3️⃣ Memory-augmented architecture: Iterative memory updates beyond the 256K window → +9.48 pts on 1M–4M token tasks! All weights, data recipes, and training code are open: 📄 Paper: https://t.co/MzlHXeardA 📥 Model: https://t.co/kE3q4HLhdk 💻 GitHub: https://t.co/KghF3mp9H3

+3 more

Meet the new Qwen3-TTS lineup: VoiceDesign & VoiceClone! 🎙️ ✨ VoiceDesign (VD-Flash): · Total Control: Design unique voices via text instructions (tone, rhythm, persona). No more presets. · Top Performance: Outperforms GPT-4o-mini & Gemini-2.5-pro on role-play benchmarks. ⚡ VoiceClone (VC-Flash): · Instant Cloning: Clone any voice from just 3 seconds of audio. · Multilingual: Supports 10 languages with 15% lower error rate than ElevenLabs.

This release was only possible by the numerous MedARC volunteers who implemented benchmarks and datasets to evaluate with. Grateful to all those who contributed! https://t.co/NwiLCJRbur

hers. #อิงฟ้ามหาชน #ชาล็อตออสติน #อิงล็อต https://t.co/WbTk6BoJ6B

+4 more

Simpsons has a bazillion legendary sign and small business gags but this King of the Hill location is so believable it hurts https://t.co/BRTbS3BDTe

are you having a laugh https://t.co/7ztqzN8qJY

Gerry Adams was posting like Chuck Grassley in 2016 at age 68. he’s now retired and Chuck is 92 and president pro tempore of the U.S. Senate https://t.co/8FnNMo7KgO

@Osse_Lives https://t.co/gbbog7vScS

Michael Clayton heads when they recognize the patina of shit coating them https://t.co/MYHFkHALbQ

Michael Clayton heads when they recognize the patina of shit coating them https://t.co/MYHFkHALbQ

we got a true reader here https://t.co/DWQ8X388xL

RACHEL'S STAYING HOME 🙌 https://t.co/BVaOkcndoD

Tim Cook’s ‘top priority’ product could finally take shape next year https://t.co/zjbX8Zj6dI by @iryantldr

Surprise! Dave Davies of the Kinks lives in New Jersey: https://t.co/axjjkJOBVC https://t.co/4uPJluzGl1

Surprise! Dave Davies of the Kinks lives in New Jersey: https://t.co/axjjkJOBVC https://t.co/4uPJluzGl1

Today, SpaceX shared a new picture (on left) of its next-generation Raptor 3 rocket engine, the most advanced rocket engine ever made. • Almost 2x the thrust of Raptor 1 • Costs 4x less • Much lighter. Will save 2,425 lbs of weight per engine, or 94,575 lbs (42.9 metric tons) per launch • No heat shield • Optimized for manufacturability It will first fly aboard Starship Version 3, with the first launch currently targeted for Q1 2026. (New photo on left from SpaceX's new quarterly impact report)

BREAKING: Grok is now integrated into 𝕏 Chat. Long press any message & select "Ask Grok". It uses an unencrypted copy of the message for analysis. Your chats are still private & encrypted. Only the message you choose is shared with Grok. Update your 𝕏 app from the AppStore. https://t.co/ttvCdXMHda

Traditional robots are like player pianos—they only play notes they’re given. Memo is more like a Master Chef who has watched hundreds of cooks in every type of kitchen. Our diverse dataset provides Memo with an “intuition” for grasping entirely new objects. This is a small step towards truly helpful robots that work in any home 🤖👨‍🍳

We introduce 🖐️🌏 𝐃𝐞𝐱𝐭𝐞𝐫𝐨𝐮𝐬 𝐖𝐨𝐫𝐥𝐝 𝐌𝐨𝐝𝐞𝐥𝐬 (𝐃𝐖𝐌) — a scene-action-conditioned video diffusion model that simulates human manipulation in static 3D scenes from egocentric hand motions. 📄 Paper: https://t.co/u2IldpfYTD 🌐 Project Page: https://t.co/YSE3rOmV1s https://t.co/mYrjUdO5aJ

🚨 BREAKING: FDA grants *Breakthrough Device* status to Paradromics’ next-gen high-bandwidth BCI system - speeding regulatory review and clinical trial support for restoring communication in paralysis. This brings real clinical neural interfaces closer to everyday impact. High-bandwidth BCIs that can stream robust neural data are foundational for future full-sensory VR and cognitive augmentation interfaces.

I"VE CLONE GEMINI AN I - https://t.co/mAvvndiL3c

https://t.co/Mdtyy1y2Ip

https://t.co/Mdtyy1y2Ip

BREAKING: SpaceX has announced that @Starlink now has over 9 million customers, up from 8M in November and 7M in August 2025. Starlink added a record 21,275 new customers on average per day since they hit 8M, smashing their previous record of 14,250. That growth rate is a whopping 49.3% higher than it was just last month! 🤯

Portraits of Iryna Zarutska went up in D.C., NYC, Miami, and Venice, Ca after Elon Musk and Eoghan McCabe pledged to fund the murals. Promises made, promises kept. Thank you @elonmusk and @eoghan. America will never forget you, Iryna. ❤️ https://t.co/ZYEPxMhIUi

+2 more

ACLU attorney in press release: “Children will die without sex-rejecting procedures!” ACLU attorney before SCOTUS, where she could get in trouble for lying: “Yeah, there is no evidence these procedures save lives.” https://t.co/9luvBNwin0

dear santa, define good 💭 https://t.co/jXZyxdIVGz