bartleby:1b, a small Large Language Model that would prefer not to

There are now two different versions of bartleby:1b:

A fine-tuned open-weight model that can be run locally and privately

This open-weight model is fine-tuned from Google’s gemma3:1b open-weight model using unsloth. This model is 806MB in size and needs about 1GB of RAM to run on a CPU (no GPU required), meaning it can run locally on ~10 year old laptops or ~5 year old smartphones. When running locally, it typically consumes as much energy when used as that device consumes when streaming video or playing video games.

To chat with it:

• Easiest way is the bartleby.stu.lu cloud-hosted (via HuggingFace) app. There is nothing to install and loads near-instantly, but runs in various commercial cloud data centers. Privacy is not guaranteed: my chat app does not log conversations, but you can’t trust me or the cloud providers!
• On desktops/laptops (Windows, Mac OS X, or Linux), you can download the ollama app for running open-weight LLMs. Pull the model stuu/bartleby:1b.
• On mobile, you can download PocketPal (iOS, Android) then go to Models -> Add Model (+ icon) -> Add from Hugging Face -> search for bartleby and select the repo staeiou/bartleby-1b.

An instruction-trained custom GPT that runs on OpenAI’s ChatGPT models

A custom GPT on openai.com, which uses OpenAI’s existing ChatGPT models. Custom GPTs work by sending an extended set of instructions and examples before your prompt to the same base ChatGPT models. The custom GPT will still answer some questions correctly, as the base models are trained to be helpful — unlike Bartleby. An OpenAI account is required, but a paid subscription not required.