20+ models
The IBM Granite Embedding 30M and 278M models models are text-only dense biencoder embedding models, with 30M available in English only and 278M serving multilingual use cases.
A general-purpose model ranging from 3 billion parameters to 70 billion, suitable for entry-level hardware.
Building upon Mistral Small 3, Mistral Small 3.1 (2503) adds state-of-the-art vision understanding and enhances long context capabilities up to 128k tokens without compromising text performance.
A family of open-source models trained on a wide variety of data, surpassing ChatGPT on various benchmarks. Updated to version 3.5-0106.
Ling-3.0-tiny is a lightweight 7.9B-parameter hybrid reasoning MoE with just 1.3B active parameters per token, offering strong reasoning and agentic capabilities at low inference cost.
https://huggingface.co/tencent/Hy-MT2-30B-A3B
Qwen3-Coder is a highly capable coding assistant, based on Qwen3-Coder-30B-A3B-Instruct. She is designed for local development and agentic coding workflows
5 Sovereign AI Models — Edge (1.5B), Reasoning (30B), Professional (22B), Verification (7B), Vision (8B). Unified ecosystem by NJIR.AI.
tongyi DeepResearch, an agentic large language model featuring 30 billion total parameters, with only 3 billion activated per token.
Fast local coding agent on LFM2 24B MoE. 30B-class quality at 8B-class speed. Tool-calling ready, 32k context. Tuned for Aider, Cline, Continue, Cursor, Codex, OpenCode, and Claude Code workflows.
Embedding Model embeddinggemma-300M-NVFP4-Q8-GGUF
Kikico's 2nd AI model, featuring 146M parameters trained only on a RTX 3080Ti
Quantized version of THUDM/GLM-4-32B-0414 optimized for tool usage with Cline / Roo Code and complex problem solving.
Fine-tuned manim model using Qwen2.5-Coder-14B-Instruct trained on 3blue1brown-manim dataset with 2,407 examples.
s1 is a reasoning model finetuned from Qwen2.5-32B-Instruct on just 1,000 examples. It matches o1-preview & exhibits test-time scaling via budget forcing.
a reasoning model finetuned from Qwen2.5-32B-Instruct on just 1,000 examples. It matches o1-preview & exhibits test-time scaling via budget forcing. available in [F16, q8_0, q6_K, q4_K_S]
from gmonsoon/MiniCPM-3B-Turangga-v3-ep50-GGUF
Q3 quantized variant of Muse Glimmer 30B for a slightly smaller footprint.
Imatrix-quantized GGUF builds of Meta's Muse Glimmer 30B (agentic, multimodal) down to 2-bit for 8–12GB setups — plus the official 17GB base build.
Granite-4.0-H-350M is a lightweight instruct model finetuned from Granite-4.0-H-350M-Base using a combination of open-source instruction datasets with permissive license and internally collected synthetic datasets.