OpenAI Go Reference Go Report Card License: MPL 2.0

Installation

$ go install github.com/picatz/openai/cmd/openai@latest

Important

To use the CLI you must have a valid OPENAI_API_KEY environment variable set. You can get one here.

Tip

You can customize which model is used by setting the OPENAI_MODEL environment variable. The default is gpt-4o today, but it may change in the future.

Usage

$ openai --help
OpenAI CLI
Usage:
 openai [flags]
 openai [command]
Available Commands:
 assistant Start an interactive assistant chat session
 chat Chat with the OpenAI API
 completion Generate the autocompletion script for the specified shell
 help Help about any command
 image Generate an image with DALL·E
 responses Manage the OpenAI Responses API
Flags:
 -h, --help help for openai
Use "openai [command] --help" for more information about a command.

$ openai assistant --help
Interact with the OpenAI API using the assistant API.
This can be used to create a temporary assistant, or interact with an existing assistant.
Usage:
 openai assistant [flags]
 openai assistant [command]
Examples:
 $ openai assistant # create a temporary assistant and start chatting
 $ openai assistant chat # same as above
 $ openai assistant create --name "Example" --model "gpt-4-turbo-preview" --description "..." --instructions "..." --code-interpreter --retrieval
 $ openai assistant list
 $ openai assistant info <assistant-id>
 $ openai assistant chat <assistant-id>
 $ openai assistant delete <assistant-id>
Available Commands:
 chat Start an interactive assistant chat session
 create Create an assistant
 delete Delete an assistant
 file Manage assistant files
 info Get information about an assistant
 list List assistants
 update Update an assistant
Flags:
 -h, --help help for assistant
Use "openai assistant [command] --help" for more information about a command.

Tip

If provided no arguments, the CLI will default to the assistant command with an ephemeral session, meaning messages and files will be deleted after exiting the session.

With Ollama

$ brew install ollama
...
$ ollama serve &
...
$ ollama run granite3.1-dense:2b
...
$ OPENAI_MODEL="granite3.1-dense:2b" OPENAI_API_URL="http://localhost:11434/v1/" openai chat