Skip to main content

​
Synopsis

ollama launch
ollama launch [INTEGRATION] [OPTIONS]
ollama launch [INTEGRATION] -- [EXTRA_ARGS...]

​
Description

The launch command provides a convenient way to:
  • Open the Ollama interactive menu
  • Launch editor integrations (Claude Code, Cline, Codex, Droid, OpenCode, OpenClaw, Pi)
  • Configure model selections for integrations
  • Pass additional arguments to launched applications
Without arguments, launch is equivalent to running ollama directly and opens the main TUI menu.

​
Arguments

​
INTEGRATION
string
Optional integration name to launch directly. Available integrations:
  • claude - Claude Code
  • cline - Cline
  • codex - Codex
  • droid - Droid
  • opencode - OpenCode
  • openclaw - OpenClaw (aliases: clawdbot, moltbot)
  • pi - Pi
​
EXTRA_ARGS
string[]
Additional arguments to pass to the integration. Use -- to separate Ollama arguments from integration arguments.
ollama launch codex -- -p myprofile

​
Options

​
--model
string
Specify a model to use with the integration without opening the selector
ollama launch claude --model llama3.2
​
--config
boolean
default:"false"
Configure the integration (select/change models) without launching
ollama launch droid --config

​
Examples

​
Launch Interactive Menu

Open the main Ollama menu: 
ollama launch
This shows:
  • Run a model
  • Launch editor integrations
  • Configure settings

​
Launch Specific Integration

Launch an integration directly: 
ollama launch claude
If not configured, you’ll be prompted to select a model.

​
Launch with Model Selection

Specify a model without prompting: 
ollama launch opencode --model llama3.2

​
Configure Integration

Change model selection without launching: 
ollama launch codex --config
This opens the model selector and saves your choice.

​
Pass Additional Arguments

Pass arguments to the integration: 
# Pass profile argument to Codex
ollama launch codex -- -p myprofile

# Pass multiple arguments
ollama launch codex -- --sandbox workspace-write --debug

# Pass arguments with model selection
ollama launch codex --model llama3.2 -- -p production

​
Supported Integrations

​
Claude Code

ollama launch claude
ollama launch claude --model llama3.2
Launches the Claude Code extension with Ollama backend.

​
Cline

ollama launch cline
ollama launch cline --config
Launches Cline with configured model(s).

​
Codex

ollama launch codex
ollama launch codex -- -p myprofile
Launches Codex with optional profile.

​
Droid

ollama launch droid
ollama launch droid --model mistral
Launches Droid with Ollama integration.

​
OpenCode

ollama launch opencode
ollama launch opencode --model codellama
Launches OpenCode with the specified model.

​
OpenClaw

ollama launch openclaw
ollama launch clawdbot  # Alias
ollama launch moltbot   # Alias
Launches OpenClaw (Clawdbot/Moltbot).

​
Pi

ollama launch pi
ollama launch pi --config
Launches Pi with model configuration.

​
Configuration

Integrations are configured in ~/.ollama/config.json: 
{
  "integrations": {
    "claude": {
      "models": ["llama3.2:latest"],
      "onboarded": true
    },
    "codex": {
      "models": ["codellama:7b", "mistral:7b"],
      "onboarded": true
    }
  },
  "last_model": "llama3.2:latest",
  "last_selection": "claude"
}

​
Configuration Options

  • models: Array of model names configured for the integration
  • onboarded: Whether initial setup has been completed
  • aliases: Custom aliases for models (integration-specific)

​
Integration States

​
Not Configured

First launch prompts for model selection: 
ollama launch claude
# Shows model picker
# Saves selection
# Launches Claude Code

​
Configured

Subsequent launches use saved model: 
ollama launch claude
# Uses saved model
# Launches immediately

​
Reconfigure

Change the configured model: 
ollama launch claude --config
# Opens model picker
# Updates configuration
# Does NOT launch

​
Model Selection

When prompted, you can select from:
  • Local models (from ollama list)
  • Cloud models (if enabled)
  • Recently used models (highlighted)
Use arrow keys to navigate and Enter to select.

​
Environment Variables

​
OLLAMA_HOST
string
default:"http://127.0.0.1:11434"
Ollama server address

​
Exit Codes

  • 0 - Success
  • 1 - Error (integration not found, model not available, etc.)

​
Troubleshooting

​
Integration Not Found

Error: unknown integration: myapp
Solution: Use one of the supported integration names: 
ollama launch  # See list in menu

​
Integration Not Installed

If the integration application isn’t installed: 
Error: integration 'claude' is not installed
Solution: Install the integration first, then try again.

​
Model Not Available

If the specified model doesn’t exist: 
Error: model 'mymodel' not found
Solution: Pull the model or choose an available one: 
ollama list
ollama pull llama3.2
ollama launch claude --model llama3.2

​
Server Not Running

Error: could not connect to ollama server
Solution: The launch command starts the server automatically if needed. If this error persists, try: 
ollama serve

​
Advanced Usage

​
Automation

Launch integrations from scripts: 
#!/bin/bash
# Launch OpenCode with specific model
ollama launch opencode --model codellama:7b

​
Multiple Configurations

Configure different models for different integrations: 
# Configure Codex for code generation
ollama launch codex --model codellama:7b

# Configure Claude for general assistance
ollama launch claude --model llama3.2

# Configure Droid for mobile development
ollama launch droid --model mistral:7b

​
Desktop Integration

Create desktop shortcuts: 
# Linux .desktop file
[Desktop Entry]
Name=Ollama Claude
Exec=ollama launch claude
Icon=ollama
Type=Application

​
Integration Architecture

The launch command:
  1. Checks server: Ensures Ollama server is running
  2. Validates integration: Confirms integration exists and is installed
  3. Loads config: Reads ~/.ollama/config.json
  4. Handles model selection: Prompts if needed or uses specified model
  5. Saves configuration: Updates config file with selections
  6. Launches application: Starts the integration with appropriate settings