We're excited to introduce Arcee.ai's Trinity Large model. An... open 400B parameter Mixture of Experts model, delivering frontier-level performance with only 13B active parameters. Trained in collaboration between Arcee, Datology and Prime Intellect.show more

We're excited to introduce Ocelot, an open-source and Brave-trained... AI model built specifically to summarize web content. Ocelot is now available in Leo, our browser's AI assistant, and as an open-source model for the developer community.show more

We are excited to introduce Mercury, the first commercial-grade... diffusion large language model (dLLM)! dLLMs push the frontier of intelligence and speed with parallel, coarse-to-fine text generation.show more

hi all, excited to join! i'm building an expressive... mini "shoggoth" robot which will eventually be hooked up to gpt4o realtime voice. i'm currently working on the low-level policies, which are trained in a mujoco simulation with RL. to delay working on raw-pixels for now, i trained a pose-estimation model using deeplabcut and triangulate the position in 3d space using the stereo cameras. eventually, i'll use gpt4o's tool calling capabilities to activate several of these policies (closed and open loop) based on the dialog flow! captions: manual actuation of the tentacle / 3d pose estimation / target designshow more

MolmoAct2 is landing in LeRobot! Ai2's open Action Reasoning... Model combines a Molmo2-ER vision-language backbone with a flow-matching continuous action expert to predict robot action chunks from images, language instructions, and proprioceptive state. An open robot foundation model built for real-world control, with strong out-of-the-box performance and easy fine-tuning in LeRobot. Pick-and-place inference running on NVIDIA DGX Spark! Blog: Paper: Thanks to Ai2 Jiafei Duan Haoquan Fangshow more

💻 Agent mode just went local! Today, Mistral AI... released Devstral, an Apache-2.0, 24B-parameter model trained for tool calling in real-world software development environments It reached #1 on SWE-bench for open-source models, but more importantly in our real-world testing it has proven to be capable of navigating and autonomously editing codebases, while running entirely on a laptop To try it out, just add `ollama/devstral` in Continue. See below for a quickstart with ollama👇show more

We are excited to introduce Stable Fast 3D, Stability... AI’s latest breakthrough in 3D asset generation technology. This innovative model transforms a single input image into a detailed 3D asset in just 0.5 seconds, setting a new standard for speed and quality in the field of 3D reconstruction! Alongside this release, we’ve also published a technical report that highlights how we achieve fast inference speeds with reduced baked illumination and material parameters. 👾You can learn more and access the report here:show more

The human brain is truly a marvel of nature.... If you horribly reductive, and boiled it down to a language model, you'd be looking at roughly 100 trillon parameters running as a sparse MoE architecture Only about 1-5% of neurons fire at any given moment, meaning the brain "activates" maybe 1-5 trillion parameters per inference step. For context, the largest AI models we've built probably top out around 5 trillion parameters. The brain is roughly 100x larger. Even its active params at any given moment are larger than almost every model in existence today. Here's what melts my brain (pun intnended) though Your brain does all of this on about 20 watts of power, less than a dim light bulb. Training a frontier AI model consumes enough electricity to power small cities for months. Running inference across data centers pulls megawatts. Your brain runs 24/7 for 80+ years on the equivalent of a phone charger. We haven't come close to matching the brain's scale. And we're not even in the same universe when it comes to efficiency. Evolution spent 500 million yrs optimizing the most energy-efficient intelligence architecture ever known. we're trying to brute force our way there with compute and electricity. Nature is still the best engineer in the room.show more

Something NVIDIA & Google do better than anyone else... is software-hardware-system co-design, and not just optimizing hardware for current model architectures, but predicting future ones. Back in early 2022, when NVIDIA started the design process for NVL72, MoE (Mixture of Experts) models were not yet the standard, and dense models were still dominant for frontier models. However, NVIDIA's strong software-hardware co-design culture enabled them to make a calculated bet that MoEs were the future, and they built NVL72 specifically for best MoE performance per TCO (Total Cost of Ownership). Furthermore, back in 2022, disaggregated prefill and wide expert parallelism (wideEP) MoE inference optimizations hadn't been invented yet, but it turns out that these MoE inference optimizations work best on large-scale systems like NVL72. While most other AI chip companies' in-house AI labs focus on training small 5B models that mainly use data parallelism, NVIDIA and Google's in-house AI labs continuously push the boundaries of model architecture and training recipes, such as NVFP4 training. Just like Super Idol & IShowSpeed, there must be a strong partnership between software engineers and hardware engineers to deliver the best systems that maximize performance per TCO.show more

We’re excited to introduce ShinkaEvolve: An open-source framework that... evolves programs for scientific discovery with unprecedented sample-efficiency. Blog: Code: Like AlphaEvolve and its variants, our framework leverages LLMs to find state-of-the-art solutions to complex problems, but using orders of magnitude fewer resources! Many evolutionary AI systems are powerful but act like brute-force engines, burning thousands of samples to find good solutions. This makes discovery slow and expensive. We took inspiration from the efficiency of nature. ‘Shinka’ (進化) is Japanese for evolution, and we designed our system to be just as resourceful. On the classic circle packing optimization problem, ShinkaEvolve discovered a new state-of-the-art solution using only 150 samples. This is a big leap in efficiency compared to previous methods that required thousands of evaluations. We applied ShinkaEvolve to a diverse set of hard problems with real-world applications: 1/ AIME Math Reasoning: It evolved sophisticated agentic scaffolds that significantly outperform strong baselines, discovering an entire Pareto frontier of solutions trading performance for efficiency. 2/ Competitive Programming: On ALE-Bench (a benchmark for NP-Hard optimization problems), ShinkaEvolve took the best existing agent's solutions and improved them, turning a 5th place solution on one task into a 2nd place leaderboard rank in a competitive programming competition. 3/ LLM Training: We even turned ShinkaEvolve inward to improve LLMs themselves. It tackled the open challenge of designing load balancing losses for Mixture-of-Experts (MoE) models. It discovered a novel loss function that leads to better expert specialization and consistently improves model performance and perplexity. ShinkaEvolve achieves its remarkable sample-efficiency through three key innovations that work together: (1) an adaptive parent sampling strategy to balance exploration and exploitation, (2) novelty-based rejection filtering to avoid redundant work, and (3) a bandit-based LLM ensemble that dynamically picks the best model for the job. By making ShinkaEvolve open-source and highly sample-efficient, our goal is to democratize access to advanced, open-ended discovery tools. Our vision for ShinkaEvolve is to be an easy-to-use companion tool to help scientists and engineers with their daily work. We believe that building more efficient, nature-inspired systems is key to unlocking the future of AI-driven scientific research. We are excited to see what the community builds with it! Learn more in our technical report:show more

With Hunyuan3D World Model 1.0 now released and open-sourced,... we're excited to showcase the technical highlights behind this impressive innovation: ✅360° Panoramic Generation: Creates complete, immersive “world scenes”, far beyond localized views. ✅Explorable 3D Scene Generation: Generates diverse, spatially consistent 3D worlds from text/image for truly immersive exploration. ✅Interactive/Editable: Achieves separation of foreground objects, background terrain, ground, and sky, for seamless secondary editing. ✅Exportable Mesh: Generated scenes can be exported as 3D meshes for direct import into mainstream game engines and modeling software. ✅Industry-Leading SOTA Evaluation: Surpasses state-of-the-art open-source models in generation quality. As the industry's first open-source model for physical simulation and explorable world generation, Hunyuan3D World Model 1.0 aims to foster a collaborative community ecosystem with developers and enthusiasts. ✨ Try it now: 🤗 Hugging Face:show more

Super excited to visit WUJI TECH who's pushing the... frontier with their advanced dexterous robotic hand Most traditional robotic hands still rely on tendons/cables, but Wuji Hand uses direct-drive actuators embedded right in the fingers with worm gears. This solves the usual tendon problems by delivering smoother, more precise, and reliable motion with way less sim-to-real gap It also has high dexterity with 20 active degrees of freedom and the weight/size feel almost exactly like an adult human hand Got to test the newly launched Wuji Hand 2 today and was super fascinated! (watch how I handle the hand in the first video😂)show more

STEVE-1: A Generative Model for Text-to-Behavior in Minecraft paper... page: Constructing AI models that respond to text instructions is challenging, especially for sequential decision-making tasks. This work introduces an instruction-tuned Video Pretraining (VPT) model for Minecraft called STEVE-1, demonstrating that the unCLIP approach, utilized in DALL-E 2, is also effective for creating instruction-following sequential decision-making agents. STEVE-1 is trained in two steps: adapting the pretrained VPT model to follow commands in MineCLIP's latent space, then training a prior to predict latent codes from text. This allows us to finetune VPT through self-supervised behavioral cloning and hindsight relabeling, bypassing the need for costly human text annotations. By leveraging pretrained models like VPT and MineCLIP and employing best practices from text-conditioned image generation, STEVE-1 costs just $60 to train and can follow a wide range of short-horizon open-ended text and visual instructions in Minecraft. STEVE-1 sets a new bar for open-ended instruction following in Minecraft with low-level controls (mouse and keyboard) and raw pixel inputs, far outperforming previous baselines. We provide experimental evidence highlighting key factors for downstream performance, including pretraining, classifier-free guidance, and data scaling. All resources, including our model weights, training scripts, and evaluation tools are made available for further research.show more

Introducing VL-JEPA: Vision-Language Joint Embedding Predictive Architecture for streaming,... live action recognition, retrieval, VQA, and classification tasks with better performance and higher efficiency than large VLMs. • VL-JEPA is the first non-generative model that can perform general-domain vision-language tasks in real-time, built on a joint embedding predictive architecture. • We demonstrate in controlled experiments that VL-JEPA, trained with latent space embedding prediction, outperforms VLMs that rely on data space token prediction. • We show that VL-JEPA delivers significant efficiency gains over VLMs for online video streaming applications, thanks to its non-autoregressive design and native support for selective decoding. • We highlight that our VL-JEPA model, with an unified model architecture, can effectively handle a wide range of classification, retrieval, and VQA tasks at the same time. by Delong Chen (陈德龙) Mustafa Shukor Théo Moutakanni Willy Jade Lei Yu Tejaswi Kasarla Allen Bolourchi Yann LeCun Pascale Fungshow more

Current Vision-Language-Action (VLA) paradigms in autonomous driving primarily rely... on Imitation Learning (IL), which introduces inherent challenges such as distribution shift and causal confusion. Online Reinforcement Learning offers a promising pathway to address these issues through trial-and-error learning. However, applying online reinforcement learning to VLA models in autonomous driving is hindered by inefficient exploration in continuous action spaces. MindDrive, a VLA framework comprising a large language model (LLM) with two distinct sets of LoRA parameters. The one LLM serves as a Decision Expert for scenario reasoning and driving decision-making, while the other acts as an Action Expert that dynamically maps linguistic decisions into feasible trajectories. Paper Title: MindDrive: A Vision-Language-Action Model for Autonomous Driving via Project: Link:show more

Human3R: Everyone Everywhere All at Once Note: I recorded... the video from the interactive demo on their project page (linked in the comment below). Abstract (excerpt): Human3R jointly recovers global multi-person SMPL-X bodies ("everyone"), dense 3D scenes ("everywhere"), and camera trajectories in a single forward pass ("all-at-once"). Our method builds upon the 4D online reconstruction model CUT3R and uses parameter-efficient visual prompt tuning to preserve CUT3R's rich spatiotemporal priors while enabling direct readout of multiple SMPL-X bodies. Human3R is a unified model that eliminates heavy dependencies and iterative refinement. After being trained on the relatively small-scale synthetic dataset BEDLAM for just one day on one GPU, it achieves superior performance with remarkable efficiency: it reconstructs multiple humans in a one-shot manner, along with 3D scenes, in one stage, at real-time speed (15 FPS) with a low memory footprint (8 GB).show more

TESLA MODEL Y JUNIPER: THE BEST INVENTION OF AFFORDABLE... LUXURY Tesla’s top-selling SUV is getting a next-level upgrade, bringing cutting-edge tech to everyday drivers. * Glides over rough roads with new suspension that makes every drive feel smooth and easy * Smarter than ever: 360° camera, sharper sensors, and upgraded Autopilot for safer, easier driving * 0–60 in just 4.3 seconds, with an even faster Performance version coming later in 2026 * Sleeker design adds range: up to 330 miles on a single charge * Starts around $37K for base model, high-end experience without the high-end price * Deliveries expected to begin in Q1 2026 in the U.S. A smarter drive, built for any road. Source: xAI, Business Insidershow more

Run Gemma 4 26B MoE on 8GB VRAM with... 250k context at 20+ tokens/sec If you own any 8GB VRAM graphics card, stop what you are doing. Local AI just had its absolute "Holy Shit" moment for budget hardware. Yesterday, I benchmarked Unsloth Gemma 4 12B Q4_K_XL on an 8GB card. The community went wild but immediately demanded more: "Can we run a 25B+ model on budget GPUs?" Today, I’m delivering exactly that. I am running a massive 26B parameter Mixture of Experts (MoE) model locally on a standard 8GB VRAM setup with 250k full native context!. If you own an RTX 3060, 3070, 4060, or any budget GPU with 8GB of VRAM, the local AI paradigm has completely changed. The performance metrics are astonishing: - 20 tokens/sec flat decode throughput. - Stable, flat decode speed even with massive prompts. - I threw a 60k token prompt at it, and it still clocked in at 20 TPS without dropping a single frame. # What about prefill? Yes, Time To First Token (TTFT) is slightly high when swallowing massive contexts. But with a solid 200 tokens/sec prefill speed, the wait is barely noticeable and highly usable. And this is running completely without Multi Token Prediction (MTP) active. How is this possible? It’s the magic of Google's new QAT (Quantization Aware Training) quants for Gemma 4. The model weight file (unsloth gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf) is only 13.2 GB, making it the ultimate local powerhouse. # The Test Setup: CPU: Intel Core i7 RAM: 16GB System RAM GPU: NVIDIA GeForce RTX 4060 Laptop GPU (8GB VRAM) # The Secret Sauce (The -cmoe Flag) To make this work properly on any 8GB card, you must use the -cmoe (CPU MoE) flag in llama.cpp. This flag isolates the heavy MoE expert weights directly to system memory (CPU/RAM) while letting your GPU focus strictly on the Attention layers and the KV Cache. It prevents VRAM spillage and holds the throughput rock solid. # The flags: -m "gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf" -cmoe -c 248000 -v Once running, just open the UI on localhost and toggle the new reasoning lightbulb icon in the text input box to watch the model perform multi step thinking. Are you still running smaller models, or are you ready to scale up your budget local setups? Let's discuss in the repliesshow more