Large Language Models (LLMs)

These open-source LLMs can be run locally and are on GitHub:

Alibaba Qwen https://github.com/QwenLM/Qwen3

Baichuan https://github.com/baichuan-inc

DeepSeek https://github.com/deepseek-ai/DeepSeek-LLM

Deepmind Gemma series https://github.com/google-deepmind/gemma

IBM Granite https://github.com/ibm-granite

Meta Llama https://github.com/meta-llama/llama

Microsoft Phi (SLM) https://github.com/microsoft/PhiCookBook

MiniMax https://github.com/MiniMax-AI/MiniMax-01

Mistral https://github.com/mistralai/mistral-inference

Moonshot Kimi https://github.com/MoonshotAI

Yi Technology - Yi series https://github.com/01-ai

Z.AI GLM series https://github.com/zai-org

Ollama or LM Studio will run .gguf models on your local network

https://github.com/ollama/ollama

https://github.com/lmstudio-ai

"Local Model" has been added to the AI Connections dialog from eSearch 1.1.9400

This will avoid the need to use Ollama or LM Studio to run local LLMs. Local Model uses https://github.com/SciSharp/LLamaSharp rather than Llama.cpp which is used by Ollama and LM Studio.

In addition from eSearch 1.1.9400 to avoid having to use Ollama or LM Studio to download LLMs, there is an AI > Import LLMs... menu item which downloads .gguf models from . You can also migrate .gguf models that were downloaded by LM Studio into TARILIO, the file format and folder structure is the same. TARILIO can be used as the server for LLMs on your local network as well as a desktop client on other machines on the network, unlike Ollama and LM Studio which both require browser based software to be installed as clients.

Anthropics DXT Claude Code etc. https://github.com/anthropics/

Other LLM Resources

https://github.com/ibm-granite-community/granite-legal-cookbook