OpenAI CLI Cheatsheet

Q: How do I install the OpenAI CLI?

Run `pip install openai`. Requires Python 3.9+. Set your API key with `export OPENAI_API_KEY=sk-xxx`. Optional extras include `aiohttp` for async, `datalib` for data analysis with GPT-4.

Q: What models are available through the CLI?

Use `openai api models.list` to see all available models. Key models include GPT-4o (general chat), GPT-4o-mini (lightweight), GPT-5.5 (latest), DALL-E-3 (image generation), Whisper-1 (audio), and TTS-1/TTS-1-HD (speech).

Q: Can I use the OpenAI CLI with third-party APIs?

Yes. Set `OPENAI_BASE_URL` to any OpenAI-compatible endpoint, or use `openai -b `. This works with Azure OpenAI, local proxies like LiteLLM, and self-hosted OpenAI-compatible servers.

Q: How do I fine-tune a model via CLI?

First upload your training data with `openai api files.create -f data.jsonl -p fine-tune`. Then start the job: `openai api fine_tuning.jobs.create -t file-xxx -m gpt-4o-mini`. Monitor with `openai api fine_tuning.jobs.retrieve ftjob-xxx`.

Q: How do I stream responses in the terminal?

Add `--stream` to your chat completion command: `openai api chat.completions.create -m gpt-4o -g user 'Tell me a story' --stream`. The output will stream token by token in real-time.

Q: How do I use Azure OpenAI with the CLI?

Use `openai -t azure --azure-endpoint https://xxx.openai.azure.com --api-version 2024-10-01 api chat.completions.create ...` or set `OPENAI_API_TYPE=azure` and `AZURE_OPENAI_ENDPOINT` environment variables.

The OpenAI CLI is the most direct tool for interacting with the OpenAI API. Install the `openai` Python package, set your API key via `OPENAI_API_KEY`, and start using chat completions, image generation (DALL-E), audio transcription/synthesis (Whisper/TTS), embeddings, file management, and fine-tuning — all from the command line.

Updated: 2026-07-20·38 commands

Quick Start

``bash # Install pip install openai

# Set API Key export OPENAI_API_KEY=sk-proj-xxx

# List available models openai api models.list

# Basic chat openai api chat.completions.create -m gpt-4o -g user "Hello!"

# Stream a response openai api chat.completions.create -m gpt-4o -g user "Tell me a story" --stream

# Generate an image openai api images.generate -p "A cat astronaut on Mars"`

`Authentication`

The CLI authenticates via the OPENAI_API_KEY environment variable. You can also pass -k to override temporarily. For Azure OpenAI, use -t azure with Azure-specific flags.

`Chat Completions`

The core feature. Use openai api chat.completions.create with -m for model and -g for messages (can be specified multiple times for multi-turn). System prompt: use -g system 'instructions'. Streaming: add --stream.

`Image Generation (DALL-E)`

Use openai api images.generate with -p (prompt), -n (count, 1-10), and -s (size: 1024x1024, 1792x1024, etc.). Edit existing images with openai api images.edit.

`Audio (Whisper & TTS)`

Transcribe with openai api audio.transcriptions.create -f audio.mp3. Translate non-English to English with openai api audio.translations.create. Generate speech with openai api audio.speech.create --input "Hello" --voice alloy.

`Embeddings`

Create text embeddings for RAG and semantic search with openai api embeddings.create -m text-embedding-3-small -i "Your text".

`Fine-tuning`

Upload training data → create job → monitor status → use the fine-tuned model ID like any other model.

`File Management`

Files are used for fine-tuning and batch processing. Manage them with openai api files.list, files.create, files.retrieve, and files.delete`.

Startup & Modes(12)

Command	Level	Example
`openai --help` Show help information	Basic	`openai --help`
`openai -k` Specify API key (overrides env var temporarily)	Basic	`openai -k sk-xxx api models.list`
`openai -b` Specify API base URL (compatible with third-party APIs)	Intermediate	`openai -b https://api.openai.com/v1 api models.list`
`openai -o` Specify organization ID	Intermediate	`openai -o org-xxx api models.list`
`export OPENAI_API_KEY` Set API key environment variable	Basic	`export OPENAI_API_KEY=sk-proj-xxx`
`export OPENAI_BASE_URL` Set custom API endpoint (third-party compatible)	Intermediate	`export OPENAI_BASE_URL=https://api.example.com/v1`
`openai api models.list` List all available models	Basic	`openai api models.list`
`openai api models.retrieve` Retrieve detailed info about a specific model	Basic	`openai api models.retrieve gpt-4o`
`openai api moderations.create` Check content policy compliance	Intermediate	`openai api moderations.create -i 'I want to harm someone'`
`openai api completions.create` Create text completion (legacy API)	Intermediate	`openai api completions.create -m gpt-3.5-turbo-instruct -p 'Once upon a time' -M 100`
`curl -X POST https://api.openai.com/v1/chat/completions` Call Chat Completions API with curl	Intermediate	`curl -s https://api.openai.com/v1/chat/completions -H 'Content-Type: application/json' -H 'Authorization: Bearer $OPENAI_API_KEY' -d '{"model":"gpt-4o","messages":[{"role":"user","content":"Hello"}]}'`
`openai -V` Check openai CLI version	Basic	`openai -V`

Chat Commands(8)

Command	Level	Example
`openai api chat.completions.create` Create a chat completion (requires model and messages)	Basic	`openai api chat.completions.create -m gpt-4o -g user 'Hello!'`
`openai api chat.completions.create -m` Specify model for chat	Basic	`openai api chat.completions.create -m gpt-4o -g user 'What is AI?'`
`openai api chat.completions.create -g` Add multi-turn messages (use multiple times)	Basic	`openai api chat.completions.create -m gpt-4o -g user 'Hi' -g assistant 'Hello!' -g user 'How are you?'`
`openai api chat.completions.create -g system` Set system prompt (system message)	Intermediate	`openai api chat.completions.create -m gpt-4o -g system 'You are a helpful assistant' -g user 'Hello'`
`openai api chat.completions.create --stream` Stream chat completion output	Intermediate	`openai api chat.completions.create -m gpt-4o -g user 'Tell me a story' --stream`
`openai api chat.completions.create -t` Set temperature (0-2, default 1)	Intermediate	`openai api chat.completions.create -m gpt-4o -g user 'Explain quantum' -t 0.3`
`openai api chat.completions.create -M` Set maximum output tokens	Intermediate	`openai api chat.completions.create -m gpt-4o -g user 'Write essay' -M 2048`
`openai api chat.completions.create -n` Generate multiple responses to choose from	Intermediate	`openai api chat.completions.create -m gpt-4o -g user 'Solve math' -n 3`

ai-image(4)

Command	Level	Example
`openai api images.generate` Generate images from text (DALL-E)	Basic	`openai api images.generate -p 'A cat astronaut on Mars' -n 1 -s 1024x1024`
`openai api images.generate -p` Specify image prompt	Basic	`openai api images.generate -p 'Cyberpunk city at night, neon lights' -n 2`
`openai api images.generate -s` Specify image size	Basic	`openai api images.generate -p 'Mountain landscape' -s 1792x1024`
`openai api images.edit` Edit existing image (based on mask)	Expert	`openai api images.edit -i input.png -m mask.png -p 'Add a flower'`

ai-audio(3)

Command	Level	Example
`openai api audio.transcriptions.create` Transcribe audio to text (Whisper)	Basic	`openai api audio.transcriptions.create -f recording.mp3 -m whisper-1`
`openai api audio.translations.create` Translate non-English audio to English	Basic	`openai api audio.translations.create -f chinese_speech.mp3 -m whisper-1`
`openai api audio.speech.create` Convert text to speech (TTS)	Basic	`openai api audio.speech.create --input 'Hello world' --voice alloy --model tts-1`

ai-embedding(1)

Command	Level	Example
`openai api embeddings.create` Create text embeddings	Intermediate	`openai api embeddings.create -m text-embedding-3-small -i 'Hello world'`

File Management(4)

Command	Level	Example
`openai api files.list` List uploaded files	Basic	`openai api files.list`
`openai api files.create` Upload a file (for fine-tuning or batch processing)	Intermediate	`openai api files.create -f training_data.jsonl -p fine-tune`
`openai api files.retrieve` Retrieve file details	Basic	`openai api files.retrieve file-xxx`
`openai api files.delete` Delete an uploaded file	Basic	`openai api files.delete file-xxx`

ai-training(6)

Command	Level	Example
`openai api fine_tuning.jobs.create` Create a fine-tuning job	Expert	`openai api fine_tuning.jobs.create -t file-xxx -m gpt-4o-mini`
`openai api fine_tuning.jobs.list` List all fine-tuning jobs	Intermediate	`openai api fine_tuning.jobs.list`
`openai api fine_tuning.jobs.retrieve` Check fine-tuning job status	Intermediate	`openai api fine_tuning.jobs.retrieve ftjob-xxx`
`openai api fine_tuning.jobs.cancel` Cancel a running fine-tuning job	Intermediate	`openai api fine_tuning.jobs.cancel ftjob-xxx`
`openai api fine_tuning.jobs.list_events` View fine-tuning job event logs	Intermediate	`openai api fine_tuning.jobs.list_events ftjob-xxx`
`openai tools fine_tunes.prepare_data` Validate and prepare fine-tuning data file	Expert	`openai tools fine_tunes.prepare_data -f training_data.jsonl`

FAQ

How do I install the OpenAI CLI?

What models are available through the CLI?

Can I use the OpenAI CLI with third-party APIs?

How do I fine-tune a model via CLI?

How do I stream responses in the terminal?

How do I use Azure OpenAI with the CLI?

This cheatsheet is compiled from official tool documentation. Last updated: 2026-07-20.