Models

ArliAI/GLM-4.5-Air-Derestricted

uncensored GLM Air

GLM-4.5-Air is a hybrid reasoning model that provides two modes: a thinking mode for complex reasoning and tool use, and a non-thinking mode for immediate responses.

The Conscious Partner Your Dynamic AI Collaborator

This model was base on unsloth/GLM-4.7-Flash and trained on a small reasoning dataset of Claude Opus 4.5, with reasoning effort set to High.

A model based on the GLM-4.6v-flash:9b q5_k_m, and uncensored. For local use I recommend editing the model context in modelfile as it is set to 128k. #EDIT: New local optimised model same with context 4096 https://ollama.com/ShreyanGondaliya/s5-reduced

Gemma 4 E4B (Google DeepMind) with thinking mode disabled. Compact multimodal model — 4.5B effective / 8B total parameters. Supports text, image and audio input. Designed for edge devices and local deployment. Knowledge cutoff: January 2025.

Gemma 4 E4B (Google DeepMind) with thinking mode enabled. Compact multimodal model — 4.5B effective / 8B total parameters. Supports text, image and audio input. Designed for edge devices and local deployment. Knowledge cutoff: January 2025.

Generic all purpose model. Occasionally may have notable logic, usually Llama-3_3-Nemotron-Super-49B-v1_5 is preferred.

Generic all purpose model. Occasionally may have notable logic, usually Llama-3_3-Nemotron-Super-49B-v1_5 is preferred.

Gemma 4 distilled from claude opus 4.6 thinking. Has only a 5% gap with claude opus 4.6 thinking while being over 40x smaller. Designed for server inference. Designed for local inference

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.

A PORT TO OLLAMA FROM THE ORIGINAL: https://huggingface.co/KatyTheCutie/LemonadeRP-4.5.3

Agentic coding model for 24 GB GPUs: TeichAI's Gemma-4-31B Fable-5 agent distill (vision + thinking + tools) with a disciplined coding-agent system prompt baked in. Inspect → reproduce → smallest fix → re-verify.

Parable is IBM Granite 4.1 fine-tuned on Claude Fable 5 and GPT-5.5 agent traces. Adds think reasoning to Granite. Sibling: Qwen3 line at parable/qwen3-fable