CLI Usage

llms.py provides a powerful command-line interface for interacting with LLMs.

Basic Usage

Simple Queries

# Ask a question
llms "What is the capital of France?"

# Multi-line prompt
llms "Explain quantum computing
in simple terms"

Specify Model

Use specific model

llms -m gemini-2.5-pro "Write a Python function to sort a list"
llms -m grok-4 "Explain this code with humor"
llms -m qwen3-max "Translate this to Chinese"

System Prompts

Add system prompt

llms -s "You're a helpful coding assistant" "Reverse a string in Python?"

Combine with model selection

llms -m claude-sonnet-4-0 -s "You are a quantum computing expert" \
  "Explain quantum entanglement"

Multi-Modal Inputs

Images

# Analyze image
llms --image photo.jpg "What's in this image?"

# Remote image
llms --image https://example.com/photo.png "Describe this photo"

# With specific model
llms -m qwen2.5vl --image screenshot.png "Extract text from this image"

# Combined with system prompt
llms -s "You are a data analyst" --image chart.png "What trends do you see?"

Audio

# Transcribe audio
llms --audio recording.mp3 "Transcribe this audio"

# Summarize audio
llms --audio meeting.wav "Summarize this meeting"

# With specific model
llms -m gpt-4o-audio-preview --audio interview.mp3 "Extract main topics"

Files (PDFs)

# Summarize document
llms --file report.pdf "Summarize this document"

# Extract information
llms --file policy.pdf "What are the key changes?"

# With specific model
llms -m gpt-5 --file handbook.pdf "List all policies"

Custom Templates

Using Chat Templates

# Load request from JSON file
llms --chat request.json

# Override user message
llms --chat request.json "New user message"

# Override model
llms -m kimi-k2 --chat request.json

# With image
llms --chat image-request.json --image photo.jpg

# With audio
llms --chat audio-request.json --audio talk.mp3

Example Template

{
  "model": "kimi-k2",
  "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": ""}
  ],
  "temperature": 0.7,
  "max_tokens": 150
}

Advanced Options

Custom Parameters

# Set temperature and seed
llms --args "temperature=0.7&seed=111" "What is 2+2?"

# Multiple parameters
llms --args "temperature=0.5&max_completion_tokens=50" "Tell me a joke"

# Stop sequences
llms --args "stop=Two,Words" "Count to 5"

# Combine with other options
llms -m grok-4 --args "temperature=0.3" "Hello"

Parameter Types

• Floats: temperature=0.7, frequency_penalty=0.2
• Integers: max_completion_tokens=100
• Booleans: store=true, logprobs=true
• Strings: stop=one
• Lists: stop=two,words

Common Parameters

• temperature: Controls randomness (0.0 to 2.0)
• max_completion_tokens: Maximum tokens in response
• seed: For reproducible outputs
• top_p: Nucleus sampling parameter
• frequency_penalty: Penalize new tokens based on frequency
• presence_penalty: Penalize new tokens based on presence
• reasoning_effort: Reasoning effort (low, medium, high)

Raw Output

# Display full JSON response
llms --raw "What is 2+2?"

Verbose Logging

# Enable detailed logging
llms --verbose "Tell me a joke"

# Custom log prefix
llms --verbose --logprefix "[DEBUG] " "Hello world"

Configuration Management

List Providers and Models

# List all enabled providers and models
llms --list
llms ls

# List specific providers
llms ls ollama
llms ls google anthropic

Enable/Disable Providers

# Enable providers
llms --enable openrouter
llms --enable anthropic google_free groq

# Disable providers
llms --disable ollama
llms --disable openai anthropic

Set Default Model

# Set default model
llms --default grok-4

# Verify
llms "Hello"  # Uses grok-4

Check Providers

# Check all models for a provider
llms --check groq

# Check specific models
llms --check groq kimi-k2 llama4:400b gpt-oss:120b

This tests:

• Provider reachability
• API key validity
• Response times
• Model availability

Custom Config File

# Use custom config
llms --config /path/to/config.json "Hello"

Initialize Config

# Create default config
llms --init

Output Formatting

Pipe to Markdown Renderer

# Beautiful terminal output with glow
llms "Explain quantum computing" | glow

Save to File

# Save response to file
llms "Write a Python script" > script.py

# Append to file
llms "Add error handling" >> script.py

Examples

Code Generation

llms -m claude-sonnet-4-0 "Python function to calculate fibonacci numbers"

Translation

llms -m qwen3-max "Translate 'Hello, how are you?' to Chinese"

Data Analysis

llms -m gemini-2.5-pro --image chart.png "Analyze this sales chart"

Document Summarization

llms -m gpt-5 --file report.pdf "Provide a 3-paragraph summary"

Audio Transcription

llms -m gpt-4o-audio-preview --audio meeting.mp3 "Transcribe and summarize"

Next Steps

• Image Support - Use images with vision models
• Audio Support - Process audio files
• File Support - Work with PDFs and documents
• Server Mode - Run as API server
• Configuration - Customize settings