Alex

Gemma 4 12B can now run locally on just 8GB RAM via Dynamic GGUFs.

Google's new model, Gemma 4 12B Unified supports image, audio and 256K context.
You can run and train the model via Unsloth Studio.

GGUF: unsloth/gemma-4-12b-it-GGUF
Guide: https://unsloth.ai/docs/models/gemma-4

5 replies

upvoted a collection about 2 months ago

olmOCR

Collection

olmOCR is a document recognition pipeline for efficiently converting documents into plain text. olmocr.allenai.org • 12 items • Updated Dec 23, 2025 • 155

New activity in unsloth/gemma-4-26B-A4B-it-GGUF 3 months ago

problem with tokenizer?

#18 opened 3 months ago by

AlexPoto

liked 4 models 4 months ago

liked a model 5 months ago

unsloth/GLM-4.7-Flash-REAP-23B-A3B-GGUF

Text Generation • 23B • Updated Jan 28 • 32.9k • 252

liked a dataset 5 months ago

nyuuzyou/suno

Viewer • Updated Feb 3 • 660k • 415 • 146

liked a model 5 months ago

speechbrain/spkrec-xvect-voxceleb

Audio Classification • Updated Feb 25, 2024 • 12.2k • 66

liked a Space 5 months ago

Qwen3-TTS Demo

🎙

2.02k

Generate speech from text using voice design, cloning or presets

liked 2 models 6 months ago

Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign

Text-to-Speech • 2B • Updated Jan 29 • 687k • 370

zai-org/GLM-4.7-Flash

Text Generation • 31B • Updated Jan 29 • 2.7M • • 1.77k

reacted to danielhanchen's post with 🔥 7 months ago

Post

4170

You can now run GLM-4.7, the new 355B parameter SOTA model on your local device (128GB RAM).✨

The model achieves SOTA performance on coding, agentic and chat benchmarks.

GGUF: unsloth/GLM-4.7-GGUF
Guide: https://docs.unsloth.ai/models/glm-4.7

3 replies

replied to Kseniase's post 7 months ago

Prompt Engineering and Context Engineering are complementary sides of the same LLM interaction process: the former crafts efficient input phrasing (like Zero-shot or Chain-of-Thought) that's cheaper and sufficient for most tasks due to minimal token usage, while the latter builds richer data environments for complex scenarios. Start with cost-effective Prompt Engineering for 80% of needs, escalating to Context Engineering only when extended knowledge or tools are required, as in your LM-Studio caching optimizations.