IBM just released their Granite 4.0 family of models: new models trained for tool calling and RAG, Apache 2.0. • Micro: transformer, dense 3B https://lnkd.in/gPRuhkKg • Micro H: hybrid, dense 3B https://lnkd.in/gTVbj3aG • Tiny H: hybrid, MoE 7B/1B active https://lnkd.in/gdE4TFpU • Small H: hybrid, MoE 32B/9B active https://lnkd.in/ghe6xnQd Trained for tool calling and RAG. Suitable for further fine-tuning. Requires LM Studio 0.3.28!
About us
Download and run local LLMs on your computer 👾 https://lmstudio.ai/download
- Website
-
https://lmstudio.ai
External link for LM Studio
- Industry
- Technology, Information and Internet
- Company size
- 2-10 employees
- Headquarters
- Brooklyn
- Type
- Privately Held
- Founded
- 2023
Locations
-
Primary
Get directions
Brooklyn, US
Updates
-
LM Studio reposted this
LM Studio is hiring an Applied AI Engineer in NYC with a focus on Agents for productivity tasks at home and at work. In-person in NYC • ship features to millions of users worldwide • research + eng + product • push the frontier of what's possible to do with open source AI. If you're in SF or elsewhere in the US, we'll relocate you to NYC. Apply here: https://lnkd.in/eXzMtdyV
-
LM Studio now has Find in Chat and Search All Chats! Find: ⌘/Ctrl + F Search: ⌘/Ctrl + ⇧ + F Update to 0.3.27 to get it 👾 https://lnkd.in/ewr6Jt7p
-
mistralai/magistral-small-2509 > New 24B reasoning model from Mistral AI > Supports 🏞️ image input and 🛠️ tool calling > Available in both GGUF and MLX in LM Studio! Model page: https://lnkd.in/eJ9gUrVh
-
There's a new open embedding model in town! Meet EmbeddingGemma. Utilize EmbeddingGemma via LM Studio's Embeddings APIs. > OpenAI compat: POST /v1/embeddings > lmstudio-python: model.embed(...) > lmstudio-js: model.embed(...) Docs: https://lnkd.in/g_TDVNBQ To download, run: lms get google/embedding-gemma-300m
Announcing EmbeddingGemma: our new open, state-of-the-art embedding model designed for on-device AI 📱 https://goo.gle/3I38560 EmbeddingGemma is the highest ranking open multilingual text embedding model under 500M on the MTEB leaderboard, and can work together with models like Gemma 3n to power advanced generative AI experiences and RAG pipelines. The small, fast, and efficient model leverages QAT to reduce RAM usage, features a 2K token context window, is trained across 100+ languages, and allows devs to build powerful apps that run directly on-device. EmbeddingGemma is offline-first by design and works with your favorite open source tools → https://goo.gle/3I4xiNp Download models on Hugging Face and Kaggle → https://goo.gle/3JSuR13 Dive into our docs → https://goo.gle/4mJOqHp And start building on Docker, Inc, llama.cpp, LM Studio, and Ollama → https://goo.gle/4nzKAAX
-
MXFP4 support for OpenAI's gpt-oss just landed in LM Studio's MLX engine! gpt-oss is natively MXFP4. This change brings in noticeable performance improvements as well. All thanks to Awni Hannun and the Apple MLX team! 🚀
-
-
Run google/gemma-3-270m in LM Studio! > 🤏 Tiny model > 💨 Extremely fast > 📝 Trained for instruction following Model page: https://lnkd.in/eyEmEymC
Introducing Gemma 3 270M! 🚀 It sets a new standard for instruction-following in compact models, while being ultra-efficient for specialized tasks: https://goo.gle/3Ja44x7 The models is built for focused tasks where efficiency is everything: 🔹 Massive 256k vocabulary for expert fine-tuning 🔹 Extreme energy efficiency for on-device AI 🔹 Production-ready with INT4 Quantization Aware Training Available now from Hugging Face, Kaggle, LM Studio, Docker, Inc, Ollama, llama.cpp, Unsloth AI, MLX, Gemma.cpp, and JAX.
-