Installing the command-line executable

Assuming you have Rust/Cargo installed, run this command in a terminal:

cargo install gpto

It will make the gpto command available in your PATH if you've allowed the PATH to be modified when installing Rust. cargo uninstall gpto uninstalls.

Back to the crate overview.

Readme

GPTO (Geppetto)

An Unofficial OpenAI Terminal Client

> gpto -h

A tiny unofficial OpenAI client

Usage: gpto [OPTIONS] <COMMAND>

Commands:
  prompt        The prompt(s) to generate completions for. Also accepts text from stdin
  conversation
  help          Print this message or the help of the given subcommand(s)

Options:
  -d, --disable-spinner            Disable the spinner and message when querying
  -s, --suffix <SUFFIX>            Text to be appended to end of response [default: ]
  -m, --model <MODEL>              Text to be appended to end of response, defaults to gpt-3.5-turbo and can be set in config
  -c, --config <CONFIG>            Absolute path of configuration. Defaults to $XDG_CONFIG_HOME/gpto.cfg
  -e, --endpoint <ENDPOINT>        URL to be queried, defaults to https://api.openai.com and can be set in config
  -t, --temperature <TEMPERATURE>  What sampling temperature to use. Higher values means the model will take more risks. Try 0.9 for more creative applications, and 0 (argmax sampling) for ones with a well-defined answer [default: 1]
  -n, --number <NUMBER>            How many completions to generate for each prompt [default: 1]
  -a, --max-tokens <MAX_TOKENS>    Maximum number of tokens to use for each request [default: 1000]
  -i, --timeout <TIMEOUT>          Maximum length of time in seconds to wait for an API request to complete
  -o, --top-p <TOP_P>              An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. We generally recommend altering this or temperature but not both [default: 1]
  -h, --help                       Print help
  -V, --version                    Print version

Learn more about how to use text completion

Install from Crates.io

Install Rust

# Linux and MacOS
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

Install GTPO

cargo install gtpo

Install from AUR

# Use yay or another AUR helper
yay gpto-bin

Install from GitHub

Install Rust

Clone the project

git clone git@github.com:alanvardy/gpto.git
cd gpto
./test.sh # run the tests
cargo build --release

You can then find the binary in /target/release/

Usage

Get a completion with default parameters

> gpto prompt --text "tell me a joke"

Q: What did the fish say when it hit the wall?
A: Dam!

Get completions using text from stdin (without displaying the spinner)

> echo "what is one plus one" | gpto prompt -d

Two

Get a completion with a different model (this example uses the leading code completion model). And yes, the generated code is not idiomatic!

Read more about models here. This CLI app uses the /v1/chat/completions endpoint.

> gpto -m gpt-4 prompt -t language is elixir\nwrite a function that raises an error if the argument is not an integer and multiplies it by 2 if it is an integer

def multiply_by_two(x)
  raise ArgumentError, "Argument is not an integer" unless x.is_a? Integer
  x * 2
end

Using a local LLM

You can run a local LLM using ollama and connect to with GPTO.

Get the model you want

ollama pull llama3

Start the ollama server

ollama serve

And then in another terminal run GPTO

gpto -m llama3 -e http://localhost:11434 prompt -t "Hello Llama"

Setting timeout

Timeout is 30s by default, this can be altered by changing the timeout option in gpto.cfg