The future of software In the limit, of course, software won't mean anything anymore. Everything digital will be instantly created at will. But in the interim period we'll wander through an increasingly odd in-between age for software. This transition periodwill be defined by three new precepts. Everflux software - instead of stepping from release to release, months apart, software now evolves continuously. Software creation has broken the sound barrier of our own keystrokes, and every model release pushes it faster. There is no reason to wait weeks or months for the next release when a single company merges over 30,000 code changes a day. Cyber capabilities are also progressing at an increasing pace. Bug and vulnerability fixes need to ship quickly and continuously. And every piece of software sits in a lattice of dependencies, each one now shipping at a blistering pace. Software will have to move accordingly fast just to keep from breaking as everything around it shifts. Translucent sources – A new territory has opened between open source and closed source. Today Claude will happily pick apart any Electron app, and any binary shipped with debug symbols has become a reverse-engineering playground for LLMs. In other words, much so-called closed-source software is now open source in all but name, and keeping code closed makes less and less sense. Even the notion of IP and copyright is breaking in software: Apps don't even need to be accessed to be reproduced. I'm writing this text on a personal text editor I prompted Claude to vibe-code from a few videos of people demoing editor features they liked. A mere description of the interface is increasingly enough for an LLM to rebuild the internals, whether or not they resemble the original. Software IP seems to be dissolving like cotton candy in a raccoon's paws. Bespoke features - Now that all software is open source, nothing prevents us from modifying everything. And personalization is just one prompt and a couple million tokens away, so there is absolutely no reason anymore to wait for someone else to add this feature you want. Everyone becomes their own product manager. Want a search box, an API bridge, a custom button? It's a couple of minutes away, mostly executed in the background of your daily life. Soon we'll want to personalize everything. A major challenge appears when bespoke features crash into our first tenet: everflux software. How do you customize software without cutting yourself off from upstream? I could fork it for good to add my search box, but I'd lose all future benefit from the software creator, be it new security updates or other updates to the rest of the software. How do you personalize software that was never meant to be personalized, without shouldering all of its future maintenance? Merge conflicts won't be the exception anymore; they'll be a constant of everflux software. Agents will rebase our personal features onto each new release overnight, and software will look less like a product and more like a living species, evolving through constant mutation. Image: M.C. Escher, Metamorphosis II, November 1939 - March 1940

18 of the 20 best open models today are Mixture-of-Experts. Dense is basically over. And yet the know-how to train a good MoE still lives mostly in the heads of a handful of teams. So we spent 800k GPU hours running the experiments ourselves and wrote everything down. Super happy to finally release "The Ultimate MoE Training Handbook", free and open, organized around the questions training teams hit: - how many experts, how big, shared or not? - why is half my step spent in all-to-all, and how do I hide it? - why is my run 25x slower with no error in the logs? - softmax, sigmoid or sqrt(softplus) router, and why keep it in fp32? - aux loss or DeepSeek's loss-free bias? - upcycle a dense checkpoint or train from scratch? - can I serve with fewer experts per token than I trained with? ⚠️ Warning: it's long (~70k words, budget a few days, sorry) but with 55 interactive figures, code you can run and numbers from our own runs. Read it here:

Reachy Mini can finally move! Today We’re launching Reachy Mini "Level Up", a set of new accessories for Reachy Mini: wheels (!), phone apps, and hats that change its personality obviously the first thing Reachy Mini did was race against Microduck (spoiler: it's faster)

My bar for AGI is still an AI winning a Nobel. But an AI researcher looking at the lm_head (that *everyone* tells you to keep in high precision) deciding to quantize it to 4 bits AND being actually right (0.7pt from bf16) is definitely impressive. Even better when the result are open-weights pushed on the Hugging Gace hub. Congrats @Recursive_SI !

A new frontier for on-device speech. We're open-sourcing Phonon, a suite of TTS models under 50m parameters, setting a new standard for naturalness and accuracy across languages, with the broadest inference support to date. Phonon-31m is available under Apache 2.0.

DeepSeek's open-source streak is getting absurd: - FlashMLA: 580 TFLOPS attention on H800
- DeepEP: MoE comms
- DeepGEMM: 1350+ FP8 TFLOPS matmul
- 3FS: they built an entire file system
- and today: the full inference engine behind V4.1-Flash, Apache 2.0 Mad respect

We found DeepSeek V4.1 Flash 1017 beats GPT 6 Sol on our multi-step offensive cybersecurity task. Sol was tested via the Daybreak trusted access program. We have been building a cybersecurity benchmark based on simulated corporate networks. This is a valuable indicator for the cyber-risk a model poses if in the hands of a bad actor. We plan to make the full evaluation public in the coming weeks but wanted to share our first immediate finding here.

The most built-upon model on the Hugging Face Hub isn’t Llama or Qwen. It’s FLUX.1-dev from Black Forest Labs, a German lab: 43k derivative models. Europe trained ~8% of the open-model compute on the Hub, but ~18% of what people build on top of it is European. Build or surrender? Build, and build open. See you at GTC Berlin this week

Hugging Face’s kickoff last night was absolute FIRE 🔥 A giant mega duck. DJs. Food, drinks, and music. Great conversations everywhere—catching up with familiar faces and making new connections. What a night. 🦆 Huge props to @jeffboudier and the @huggingface team for putting together such a stellar event! we missed you @ClementDelangue. Today, the ExecuTorch Hackathon is in full swing at Founders, Inc., with teams building across three tracks: 💻 Compute — on-device AI on Snapdragon X Elite-powered Copilot+ PCs 📱 Mobile + XR — AI experiences on Samsung Galaxy S25 Ultra and Galaxy XR 🛠️ IoT — edge AI, sensing, and control on Arduino VENTUNO Q boards From the dance floor to building on real hardware. Open Source AI Week is starting with a real bang—and we’re only getting started. 🚀 Hackathon details: Thanks to hackathon sponsors: @AIatMeta @Qualcomm @github @open_intel_fdn @fdotinc @hackersquadio @PyTorch #OpenSourceAIWeek# #OpenIntelligence#

OpenCode is currently under a sustained cyberattack, and mid-incident they ran into the same problem Hugging Face did: the frontier models they were using for defense began refusing to help fight attacking agents because analyzing the attacks started triggering safety guardrails.

300 hours. 28,993 demonstrations. 59 tasks. Touch on all 10 fingertips, in every frame. We're thrilled to share SharpaDex v1.0 with the robotics community! To our knowledge, SharpaDex is the largest publicly available real-robot dexterous-hand manipulation dataset with fingertip tactile sensing, measured by publicly released demonstration hours containing tactile data as of October 2026. This release brings together: • Diverse manipulation: tool use, assembly, deformable-object handling, and long-horizon tasks. • Synchronized sensing: multi-view RGB, fingertip tactile images, fingertip force/torque, and 65-dimensional joint actions. • Contact-rich data: approximately 195 hours with at least one fingertip in contact, including 55 hours of simultaneous contact across both hands. • Hierarchical language annotations: task descriptions, 217,642 temporally grounded subtask segments covering 89.3% of frames, and 45 distinct skill labels. What excites us most is what comes next: researchers using these data to explore how robots can learn from touch, coordinate both hands, and perform increasingly complex tasks. We hope SharpaDex becomes a useful resource for visual-tactile VLA models, tactile representation learning, and general-purpose dexterous manipulation. We can't wait to see what you build—and learn alongside you. Available in LeRobot v3.0 and v2.1 formats under CC BY-NC 4.0. Explore the data, tasks, and annotations through our report and interactive visualizations, or download the dataset to get started: Dataset on Hugging Face：

Fun and crazy napkin math: Meta, NVIDIA, Alibaba, DeepSeek, Moonshot, Mistral and ~30 other labs have shared close to $1B worth of training compute on the Hugging Face Hub Free to download End of 2023 it was under $40M Wikipedia was written in volunteer hours. This AI age commons is written in GPU-hours and data (it's a floor btw: about half comes from official compute numbers rest is educated guess and it still miss the ~900k community fine-tunes) Space with data to explore: