tgpt is a cross-platform command-line interface (CLI) tool that allows you to use AI chatbot in your Terminal without requiring API keys.
- KoboldAI (koboldcpp/HF_SPACE_Tiefighter-13B)
- Phind (Phind Model)
- Llama2 (Llama 2 70b)
- Blackbox AI (Blackbox model)
- OpenAI (All models, Requires API Key)
Image Generation Model: Craiyon V3
Usage: tgpt [Flags] [Prompt]
Flags:
-s, --shell Generate and Execute shell commands. (Experimental)
-c, --code Generate Code. (Experimental)
-q, --quiet Gives response back without loading animation
-w, --whole Gives response back as a whole text
-img, --image Generate images from text
--provider Set Provider. Detailed information has been provided below
Some additional options can be set. However not all options are supported by all providers. Not supported options will just be ignored.
--model Set Model
--key Set API Key
--temperature Set temperature
--top_p Set top_p
--max_length Set max response length
Options:
-v, --version Print version
-h, --help Print help message
-i, --interactive Start normal interactive mode
-m, --multiline Start multi-line interactive mode
-cl, --changelog See changelog of versions
-u, --update Update program
Providers:
The default provider is phind which uses Phind model.
Available providers to use: openai, opengpts, koboldai, phind, llama2, blackboxai, ollama and groq
Provider: openai
Needs API key to work and supports various models
Provider: opengpts
Uses gpt-3.5-turbo only. Do not use with sensitive data
Provider: koboldai
Uses koboldcpp/HF_SPACE_Tiefighter-13B only, answers from novels
Provider: phind
Uses Phind Model. Great for developers
Provider: llama2
Llama 2 is an open source large language model (LLM) developed by Meta AI. Uses llama2-70b by default. Supports other models.
Provider: blackboxai
Uses BlackBox model. Great for developers
Provider: ollama
Needs to be run locally. Supports many models
Provider: groq
Requires a free API Key. Supports LLaMA2-70b & Mixtral-8x7b
Examples:
tgpt "What is internet?"
tgpt -m
tgpt -s "How to update my system?"
tgpt --provider opengpts "What is 1+1"
tgpt --provider openai --key "sk-xxxx" --model "gpt-3.5-turbo" "What is 1+1"
cat install.sh | tgpt "Explain the code"
The default download location is /usr/local/bin, but you can change it in the command to use a different location. However, make sure the location is added to your PATH environment variable for easy accessibility.
You can download it with the following command:
curl -sSL https://raw.githubusercontent.com/aandrew-me/tgpt/main/install | bash -s /usr/local/binIf you are using Arch Linux, you can install the AUR package with paru:
paru -S tgpt-binOr with yay:
yay -S tgpt-binYou need to add the Go install directory to your system's shell path.
go install github.com/aandrew-me/tgpt/v2@latest-
Scoop: Package installation with Scoop can be done using the following command:
scoop install https://raw.githubusercontent.com/aandrew-me/tgpt/main/tgpt.json
If you installed the program with the installation script, you may update it with
tgpt -uIt may require admin privilages.
Support:
- Http Proxy [
https://ip:port] - Http Auth [
https://user:pass@ip:port] - Socks5 Proxy [
socks5:https://ip:port ] - Socks5 Auth [
socks5:https://user:pass@ip:port]
If you want to use a proxy, create proxy.txt file in the same directory from where you are executing the file and write your proxy configuration there.
Example:
https://127.0.0.1:8080You can download the executable for your operating system, rename it to tgpt (or any other desired name), and then execute it by typing ./tgpt while in that directory. Alternatively, you can add it to your PATH environmental variable and then execute it by simply typing tgpt.
If you installed with the install script, you can execute the following command to remove the tgpt executable
sudo rm $(which tgpt)
Configuration file is usually located in ~/.config/tgpt on GNU/Linux Systems and in "Library/Application Support/tgpt" on MacOS