Introducing Mythos – AI native Reader that actually understands what you’re reading. Highlight a passage → ask a question → get an answer, without breaking flow. Books, PDFs, research papers. All inside one app.

You can now train LLMs 3× faster with no accuracy loss, via our new RoPE and MLP kernels. Our Triton kernels plus smart auto packing delivers ~3× faster training & 30% less VRAM vs optimized FA3 setups. Train Qwen3-4B 3x faster on just 3.9GB VRAM. Blog: docs.unsloth.ai/new/3x-faster-…

The real reason Bryan is trying to live forever is because he was an early investor in Databricks and is waiting for the liquidity event

Single rollout RL for multimodal RL. It is similar to the previous approach of single rollout RL (arxiv.org/abs/2509.13232) but they were able to stabilize this only after applying advantage shaping with an entropy bonus.

اللقاء هذا كان من امتع اللقاءات الي حضرتها وأقربهم لقلبي 😍 يعطيه العافيه عثمان (osman) على المحتوى والحضور على النقاشات المثرية 🙏 ويعطيهم العافيه Dal | دال و MOZN مُزن على دعمه ❤️

في ظل التسارع الكبير في مجالات البيانات والذكاء الاصطناعي، أصبح من الصعب مواكبة أحدث التطورات والأبحاث العلمية، وفهم كيفية عمل هذه التقنيات والنماذج وبنائها، في وقت يقابِل ذلك نقصٌ واضح في المحتوى التعليمي المتخصص والعميق، خصوصًا باللغة العربية. نعلن عن تأسيس منصة ساعي التعليمية

Z.ai is set for its IPO on Jan 8, 2026. This journey has been powered by our developers, researchers, and users from Day 1. Thank you for building this reality with us!

في إطار الدعم المتواصل والمستمر من سمو ولي العهد رئيس مجلس الوزراء رئيس مجلس إدارة #سدايا - حفظه الله -، وضع حجر الأساس لمركز بيانات سدايا "هيكساجون".. الذي يُعد أكبر مركز بيانات حكومي في العالم، وفق أعلى تصنيف عالمي لمراكز البيانات (Tier IV)  #مركز_هيكساجون

"Recursive Language Models" A potentially big direction for LLMs in 2026 from MIT researchers In their approach, a prompt isn’t “run” directly, instead it’s stored as a variable in an external Python REPL, and the language model writes code to inspect/slice/decompose that long

we built the largest opt in human data collection effort in history. - 3 million files uploaded daily - 200,000+ data contributors in 2 weeks - data contributors earning up to $6,400 per month 12,000+ structured datasets created spanning egocentric data, medical data (e.g.,

We have raised a $110 billion round of funding from Amazon, NVIDIA, and SoftBank. We are grateful for the support from our partners, and have a lot of work to do to bring you the tools you deserve.

ByteDance just published something I've been waiting for someone to build: CUDA Agent! It trained a model that writes fast CUDA kernels. Not just correct ones — actually optimized ones. It beats torch.compile by 2× on simple/medium kernels, ~92% on complex ones, and even

Saudi Arabia’s biggest oil refinery has restarted operations after being halted due to a drone attack earlier this month bloomberg.com/news/articles/…

Riyadh

Saudi Arabia is in profound sorrow. Vital energy facilities in the Kingdom were recently targeted multiple times. One Saudi industrial security officer was killed and seven others were injured while on duty protecting them. May Allah have mercy on him and heal the wounded.

تشرفنا باستضافة تركي بن زرعه Turki Bin Zarah تركي بن زرعه، الشريك المؤسس والرئيس التنفيذي للعمليات لشركة تمارا، في جلسة حوارية بالتعاون مع نادي الشرق الأوسط وشمال أفريقيا في هارڤارد، والتي أدارتها عضوتنا وطالبة الماجستير نورة الحسين. مع جزيل الشكر لضيفنا الكريم على مشاركته خبراته من إحدى أبرز شركات

Claude now connects to the tools creative professionals already use. With the new Blender connector, you can debug a scene, build new tools, or batch-apply changes across every object, directly from Claude.