locopilot models

Manage models.

Source: src/cli/commands/models.ts

Subcommands:

• pull
• list
• rm

Model names are validated against ^[a-zA-Z0-9:._-]+$ to prevent shell injection — only letters, digits, colons, dots, hyphens, and underscores are allowed.

---

locopilot models pull <model>

Pull a model into the local Ollama runtime.

locopilot models pull llama3:8b
locopilot models pull qwen2.5:7b-instruct
locopilot models pull mistral:latest

This calls ollama pull with stdio: 'inherit', so you get Ollama's native progress bars in your terminal:

pulling manifest
pulling 8eeb52dfb3bb... 100% ▕███████████████▏ 4.7 GB
pulling 4fa551d4f938... 100% ▕███████████████▏  12 KB
verifying sha256 digest
writing manifest
success

  ✔ Model llama3:8b ready.

locopilot models list

List models reachable through the local API. Combines local Ollama models with the cloud catalog (when logged in).

locopilot models list

  Models:

  [local]  llama3:8b
  [local]  mistral:7b
  [remote] llama-3.1-70b-instruct
  [remote] qwen-2.5-72b-instruct

[local] models live on your machine and route to Ollama. [remote] models live in LocoPilot Cloud and require an active Pro subscription to invoke.

Backed by GET /v1/models — see API → models.

locopilot models rm <model>

Remove a local model from Ollama (frees disk space).

locopilot models rm llama3:8b

Calls ollama rm directly. If the model is in use, Ollama will refuse and the CLI surfaces the error.

Errors

Error	Cause
Invalid model name.	Name failed the allowlist regex
ollama not found.	ollama binary is not on $PATH — re-run locopilot init
API returned 502 / connection refused	The local API isn't running — locopilot start first