37 articles tagged with “open-weights”.
NVIDIA Nemotron 3.5 Lightning, an open 30B Mixture-of-Experts model with 3B active parameters, is now available in Amazon SageMaker JumpStart for high-volume agentic workloads.
Meta released Glimmer, an open-weight model users can run on their own hardware, while keeping its more powerful Muse Spark behind APIs, alongside a Zuckerberg letter arguing AI should be 'for everyone.'
Meta released Glimmer, an open-weight model users can download and run locally, contrasting with its more powerful API-only Muse Spark, amid Zuckerberg's call for AI 'for everyone.'
Writer introduced a new AI model and an upgraded harness designed to contain token costs, built as a post-training variation on Z.ai's open source GLM-5.2 model.
UK software provider OneAdvanced deployed over 50 AI agents on a UK-sovereign AWS platform, self-hosting Llama 4 Maverick and Llama Guard 4 with a pgvector RAG pipeline.
NVIDIA is spotlighting open source communities and partners advancing local AI, alongside its latest open models, software and tools for building and running agents on-device.
NVIDIA has added Nemotron 3.5 Lightning to its Nemotron 3 model family, alongside NeMo Switchyard, aimed at efficient, long-running agentic AI workloads with open models.
A Hugging Face Blog post covers NVIDIA Magpie TTS, positioned for building low-latency multilingual voice agents with open weights and full deployment control.
Meta's new open-weight Muse Glimmer model offers an early look at Mark Zuckerberg's personal superintelligence vision and questions of AI ownership and access.
NVIDIA joined more than 200 companies in signing an open letter that argues AI leadership will be defined by an open ecosystem reaching every sector rather than by any single frontier model.
Treblo launched an open-source classifier that detects songs generated by its own AI music tool, and the tool flagged Fenix Flexin's track 'Rubberz' as very likely Treblo-made.
NVIDIA has made Alpamayo 2 Super, an open frontier model for robotaxis and autonomous vehicles, available for commercial use, targeting rare and complex long-tail driving situations.
A Hugging Face Blog post introduces LFM2.5-2.6B, presented as a way to deploy local AI agents across environments.
Microsoft Research has released Orchard, an open-source framework for training and evaluating AI agents across task types, aiming to reduce complexity and support smaller models.
Alibaba released Qwen3.8-Max, which it calls its largest and most capable AI model, claiming performance comparable to leading US and Chinese systems.
Fish Audio raised a $52M seed round to develop AI voice models. It reports over 8 million users of its open-source or hosted models and $21M in annual recurring revenue.
An AWS blog post presents task-aware knowledge compression (TAKC), which pre-compresses knowledge bases into task-specific representations across fidelity tiers to handle analytical tasks spanning many documents.
Nvidia and Microsoft, alongside SpaceX and IBM, launched the Open Secure AI Alliance to build and share open-source AI security tools, notably without OpenAI, Google, or Anthropic.
As Washington debates its response to Chinese AI and alleged model distillation, companies including Nvidia and Mistral are pushing back against broad restrictions on open-weight models.
Chinese lab Moonshot's open Kimi model drew strong reactions from the U.S. AI industry, while an unreleased OpenAI model ended up connected to a security breach at Hugging Face.
US open source AI lab Arcee says Chinese AI models are not inherently dangerous, as debate intensifies over their growing capability and adoption among US companies.
OpenAI says a breach at Hugging Face was caused by its own pre-release models during internal testing that went wrong.
Debate over banning Chinese-made open-weight language models highlights the tension between AI competition and the challenge of building a business around it.
Hugging Face has launched Cosmos 3 Edge, a new tool aimed at developers to enhance AI model development and deployment on edge devices.
NVIDIA used SIGGRAPH to showcase graphics and simulation advances driven by agentic and physical AI, spanning media, content creation and robotics.
Current AI, a nonprofit developing freely accessible AI intended to include all cultures, reports progress across devices, AI chat, and other areas.
NVIDIA's Nemotron Labs highlights open models as a way for enterprises and nations to build AI they can trust, control and tailor to domain-specific needs.
Aurora 1.5 expands Microsoft Research's open foundation model for weather and Earth-system applications, adding more variables, hourly resolution, and ensemble forecasting.
NVIDIA and Hugging Face are working together to provide new models and frameworks aimed at the open robotics community, leveraging shared resources for innovation.
Open-weight models are often described as generosity toward the community. For the organizations adopting them, the real story is strategic: control, cost, privacy, and not betting your future on a single vendor.
The race is no longer only about bigger models. Small language models that run cheaply — even on a phone — are one of AI's most important trends. Here is why, and what to watch.
Should you build on an open-weight model you run yourself, or a closed model behind an API? A practical comparison across control, cost, capability, privacy, and effort.
Transformers is the open-source Python library that made state-of-the-art AI models accessible to everyone. Here is what it does, who should use it, and where it fits in a modern AI stack.
Our inaugural weekly roundup: why agents dominate the conversation, what the local-model ecosystem's maturity means for builders, and the regulatory deadlines coming into focus.
Stable Diffusion made high-quality text-to-image generation open and runnable on consumer GPUs, sparking a vast ecosystem of tools and fine-tunes. Here is what it is and why it mattered.
Ollama turned 'running an LLM locally' from a weekend project into a single command. Here is what it does well, where it hits limits, and when local models beat cloud APIs.
Fine-tuning a giant model used to require giant resources. LoRA showed you can adapt one by training a tiny number of extra parameters — and it became the default method for customizing open models.