Python (and C++ and R) project management. Built on the conda dev environment, and is primarily aimed at ML, data science, scientific python.
Comes highly recommended by others.

Run LLMs and other machine learning models locally. Wants to be a complete drop-in replacement for OpenAI so it also has the same REST API bindings - theoretically allowing you to replace any software that uses OpenAI models with this one (e.g. through redirecting DNS resolution in hosts).

Supports GPU acceleration but also runs without CPU. Usable for text, audio, video, image, voice generation.

Run LLMs on your own: software that can simply load LLMs locally and let you interface with them (for generative text, coding assistance, image generation, etc).

A whole LLM embedded in a font!

You can write something and then add a lot of exclamation marks and.. it just generates text for you.
But the text doesn't really exist (try copy pasting into different software) it essentially just represents it via the font to you.

This is, frankly, insane.

A fork of a pretty sweet ChatGPT plugin for neovim which seeks to make it work in an improved way with FOSS versions of GPT trained LLMs. Currently still unusable since it takes too long to generate output on most machines but with improvements will signify a way to use LLMs while staying FOSS and ideally also privacy-aware and offline.

Use chatbot models from your own system without having to interact with e.g. the OpenAI API.

Provides a litany of different chat bots, some having almost the same quality of output as ChatGPT3.5 (as of now).

States it does not require a GPU to work well.

Commandline ChatGPT interaction program, using golang. Has few quality of life features (but a pre-determined role to generate and execute shell commands [DANGER DANGER] or to generate code for you).
Has basic interactive mode.

Does NOT require OpenAI API keys, uses API from a different server. (For now, no key required).

Commandline ChatGPT interaction program, using python. Has a few quality of life features like conversation continuing, and token cost estimation.

Requires OpenAI API key to function.

A rusty commandline program to chat with OpenAI's ChatGPT. Requires OpenAI API key.

Has quite a few quality of life features like pre-determined roles (in yaml format), conversation saving, repl mode and more.

A shell script to interact with ChatGTP or DALL-E through CLI. Requires openai api key.
Built solely with (bash) shellscript, so no other interpreters required (python, node, ..)

Host your own ChatGPT-like AI API. With this you could plug your self-hosted version into any application (that allows plugging custom API URLs) and off you go with whatever model you chose!

I have no idea about the performance with/without GPU.

An AI assistant - not entirely sure what the difference to e.g. MyCroft is - but seems a little newer, coming along well. Can be integrated with a variety of TTS and STT services, and skills created for it

Very neat new TTS engine, works in a wide variety of languages (e.g. English UK/US, German, Afrikaans) and with different inflections and voices.

Also works locally and (once the generator is downloaded the first time) super fast (on my pc takes around 10 seconds for generating a minute of speech in German). Since you will usually have shorter speech snippets will work even faster.

Can be used in client/server setup, through the local cli or as part of the mycroft engine. Neat!

List of resources to delve deeper into data science and/or data engineering. Very interesting suggestions and enough overlap that it's not just a 'random list'

Pre-made models trained for natural language processing -- can summarize, generate, identify, answer questions, etc.

Process image, video, audio, or text data using deep learning models from the Model Asset Exchange in Node-RED flows.