Note that these models are tiny, so I would always recommend fine-tuning them for your use case. My colleague @EdoardMosca created two Colab notebooks to help you in this adventure. You can find them in the model cards. More tutorials to come! 🙂
OPEN SOURCE
-
LFM2-1.2B outperforms larger Qwen3 model through efficient training
By
–
This speed should come at a cost but, thanks to efficient training, we managed to squeeze incredible performance in these tiny models. LFM2-1.2B is competitive with Qwen3-1.7B, a model 47% larger!
-
Liquid AI Open-Sources LFM2 Edge LLMs Generation
By
–
Liquid AI open-sources a new generation of edge LLMs! I'm so happy to contribute to the open-source community with this release on @huggingface
! LFM2 is a new architecture that combines best-in-class inference speed and quality into 350M, 700M, and 1.2B models. -
Huggingface explains stateless direct response MCP server choice
By
–
If you're developing MCP servers, you should give a read to how the @huggingface team built the Hub MCP, they explain why they chose a Stateless + Direct Response server over other options!
-
Hugging Face Launches Affordable Reachy Mini Desktop Robot
By
–
Hugging Face opened orders for Reachy Mini, an expressive, open-source desktop robot
— The Rundown AI (@TheRundownAI) 10 juillet 2025
Starting at just $299 and fully programmable in Python, it makes an ideal candidate for human-robot interaction, creative coding, and AI experimentationpic.twitter.com/C5AxiZdXyAHugging Face opened orders for Reachy Mini, an expressive, open-source desktop robot Starting at just $299 and fully programmable in Python, it makes an ideal candidate for human-robot interaction, creative coding, and AI experimentation
-
Ai2 Launches FlexOlmo: Decentralized Language Model Co-Development
By
–
Ai2 dropped FlexOlmo, a new paradigm for the co-development of language models
— The Rundown AI (@TheRundownAI) 10 juillet 2025
It allows each data owner to locally branch from a shared public model, add an expert trained on their data locally, and contribute this expert back to the shared modelpic.twitter.com/xiaaBKKOTSAi2 dropped FlexOlmo, a new paradigm for the co-development of language models It allows each data owner to locally branch from a shared public model, add an expert trained on their data locally, and contribute this expert back to the shared model
-
Nostr Protocol: Decentralized Messaging Alternative Emerges
By
–
Interesting. An alternative to Signal and WhatsApp. New tech built on the Nostr protocol will keep increasing.
-
MedGemma: Small Multimodal LLM for Medical Professionals
By
–
MedGemma is a really interesting model – very small, multimodal, open, and does quite well in out-of-distribution medical tasks compared to much larger models. Would love to see more work thinking about how to improve & deploy this sort of LLM to support medical professionals
-
MedGemma: Open Weights Multimodal Model for Medical Data Analysis
By
–
Check out our state-of-the-art open weights MedGemma multimodal model for making sense of longitudinal EHR data as well as medical text and medical imaging data in various modalities (radiology, dermatology, pathology, ophthalmology, etc.) See the blog post linked below!
-
FlexOlmo: Distributed AI Model with Confidential Data Contribution
By
–
New on @WIRED
: A novel type of distributed mixture-of-experts model from Ai2 (called FlexOlmo) allows data can be contributed to a frontier model confidentially, and even revoked after the model is built: