Looking for your next AI Engineering Role? Aa bunch of VC-backed startups are hiring founding engineers, in ai and backend systems. Don't miss out on some great opportunities and your next big challenge Apply now! https://t.co/1wqJdsRLpS

Balaji Srinivasan says every startup now needs a founding creator. Someone who speaks for the product before it ships. “The founding engineer is the how… The founding creator is the why.” @balajis https://t.co/gmbgWa0PRw

working a data scientist agent arena for thanksgiving https://t.co/dSzHgSP73x

xRouter Training Cost-Aware LLMs Orchestration System via Reinforcement Learning https://t.co/I37meQjPKX

🚀 Eval Protocol is Open Sourced! Reinforcement fine-tuning is complicated, because there are hundreds of environments and tens of trainers you can pick and choose and integrate with. Even worse, in production, agents don’t live in clean “gyms.” They operate in messy, async environments - flaky APIs, partial observability, conflicting objectives, long feedback loops. We solve that problem by open sourcing Eval Protocol. The goal is for you to build your production RFT flow without reinventing the wheel of managing such complex integration. 👉 Day 0 support for trainers and environments like TRL (@huggingface), @rllm_project , OpenEnv (@PyTorch),  as well as support for proprietary trainers like @OpenAI RFT and @thinkymachines Tinker. More to come. 👉 Instrument agents in production instead of toy or simulated environments 👉 Move from offline benchmarks to live, continuous improvement

NVIDIA just released PhysicsNeMo-CFD-Ahmed-Body on Hugging Face This dataset provides rich Ahmed body surface data for aerodynamic simulations, including 3D geometry, pressure, and wall shear stress. Perfect for R&D and teaching CFD models! https://t.co/PUyk2bVqAi

FLUX.2 just landed in LTX. We’re proud to be official launch partners with @bfl_ml FLUX.2 is built for production with speed, precision, and photorealism. Here’s what makes it different 🧵 https://t.co/UykteGVbQz

Microsoft & Oxford Research ask: Are your UIs ready for agents? Introducing a breakthrough framework where Computer-Use Agents act as judges, evaluating & refining generative UIs for agent-native efficiency and task success, not just human aesthetics. https://t.co/dFQrefzgk1

Budget-Aware Tool-Use Enables Effective Agent Scaling https://t.co/KixL84JySx

discuss: https://t.co/3ZelB5s88o

PRInTS Reward Modeling for Long-Horizon Information Seeking https://t.co/9dqhO2UUZg

discuss: https://t.co/K1x0JQCLvh

Introducing: Retake For the first time ever, direct your shot after it's rendered. Rephrase dialogue, reshape emotion and reframe moments, without redoing the full video. RT + Follow for 800 credits + guide → DMs (First 500) What’s possible 🔽 https://t.co/Hz2lebWLcE

7000+ devs at MCP's Birthday Hackathon! But here's what smart participants know: The special awards have way better odds than the main tracks. - $2,500 cash from Modal, Blaxel - 1-year ChatGPT Pro from OpenAI - $30K in Google Gemini credits - AirPods 4 Pro from ElevenLabs https://t.co/ON88UnLGqY

iMontage: Unified, Dynamic Many-to-Many Image Generation This new framework repurposes powerful video models into an all-in-one image generator. It consumes & produces variable-length image sets for diverse tasks, delivering extraordinary dynamics & consistency. https://t.co/Y59TaKBsQK

MobileVLA-R1 Reinforcing Vision-Language-Action for Mobile Robots https://t.co/vE9DQGcuOs

discuss: https://t.co/WowJJLWTzg

Latent Collaboration in Multi-Agent Systems https://t.co/F8RxZ3B2HY

discuss: https://t.co/hkGFZcLaVG

Apple presents CLaRa Bridging Retrieval and Generation with Continuous Latent Reasoning https://t.co/Qil4TDa4Ld

discuss: https://t.co/R1qLr7TvCA

Thank @_akhaliq for introducing our efficient latent-based multi-agent systems (LatentMAS). Paper: https://t.co/yH9RFRmy72 Code: https://t.co/lRoMLHyPa9

Brand new image models FLUX.2 Pro and Flex are on the @yupp_ai image leaderboard! We’ve accumulated ~8K votes from users around the world for their everyday use cases. Both models sit below Nano Banana Pro but are ahead of the others! Congrats @bfl_ml! https://t.co/2YhjDbs9mQ https://t.co/sON90qlEVP

Fara-7B is our first agentic small language model for computer use. We learned a lot, and looking forward to next steps: *Agentic models can be small, yet remain capable *Unlike solutions that rely on chat model wrappers, even small agentic models can process screenshots and perform direct GUI actions such as scrolling, typing, and clicking. *Simulation-driven multi-agent synthetic data to automates task generation, trajectory generation and validation is a way to address the agentic data scarcity gap, and in our case costs < $1 per task. *Evaluating CUA is hard ; we release WebTailBench, a new eval set with diverse tasks not found in other benchmarks, and work with an external party, Browserbase, to independently assessed Fara-7B using human annotators. Model available on Foundry and HuggingFace and can run on device on Copilot+ PC

Cool, still pretty popular. @OpenMed_AI https://t.co/iBqoVSqL3B

Cool, still pretty popular. @OpenMed_AI https://t.co/iBqoVSqL3B

Combining two wonderful open source projects for fun as a side project. "SO-100" + "Amazing Hands" = SO-Hands Attn @LeRobotHF 😀 https://t.co/frTerj4GlS

Daily Papers has gained a lot of attention this past year with all the new updates 🔥 Here’s a guide that will help you quickly understand what’s new and make better use of the tool 🤗 https://t.co/kT5W7rzcQM

🤗 Inference Providers = access to the best open models via 19+ providers through a single API. Just shipped integration docs showing how to use Inference Providers with your favourite applications, tools and frameworks! https://t.co/Rlj2Wlp1RR

Our new paper Economies of Open Intelligence is out and covered by @Melissahei in the @FinancialTimes. It offers the clearest picture yet of how global power is shifting inside the open AI ecosystem, and what it means for the open-source community. https://t.co/jogLtb9AIZ

Who is winning the open AI race? Our new study "Economies of Open Intelligence" maps 2.2B @huggingface downloads across 851k models (2020→2025). 1) Power is rebalancing (US big tech ↓; China + community ↑) 2) Models got big & efficient (MoE, quant, multimodal surge) 3) Intermediaries now matter (adapters/quantizers steer usage) 4) Transparency is slipping /🧵

you gotta go fast and go read the latest blog by @remi_or_ et al. explaining Continuous Batching in depth https://t.co/DonRhtg7Gl https://t.co/c3qYDcxarD