LLM Status
Find every AI model your code calls and warn before it's retired
Last checked:
What is LLM Status?
LLM Status is a CLI tool and registry that scans code repositories to detect AI models in use, tracks model retirement dates, and automatically suggests or applies replacements for deprecated models. It's designed for developers to manage model dependencies across their codebase with GitHub integration for PR-level checks.
LLM Status pricing
Pricing model: Freemium
LLM Status pros
- Automated model scanning and dependency detection across repositories
- Live registry with 528 models across 15 providers, refreshed every 6 hours from official sources
- Automatic fixes via `mm fix` command that rewrite deprecated model IDs with boundary-safe matching
- No account required; works offline with signed registry and optional CI/CD integration
- GitHub App integration adds model-lifecycle checks to pull requests
- Multiple interfaces: interactive TUI dashboard, CLI commands, and scriptable JSON output
LLM Status cons
- Requires models to be explicitly referenced in code; can't detect models used only at runtime or in config files
- Accuracy depends on registry data from external sources (models.dev, OpenRouter, provider APIs)
- Auto-update mechanism runs in background without explicit user control
Frequently asked questions about LLM Status
Do I need an account to use LLM Status?
No. The tool is free, works offline with a signed registry, and only tracks anonymous usage counts (which can be disabled with MM_NO_ANALYTICS=1).
How does `mm fix` handle model replacement?
It rewrites retiring model IDs to registry-recommended replacements, preserving quotes and formatting. It's boundary-safe (e.g., gpt-4 won't rewrite inside gpt-4o) and chains through dying replacements to the first live model.
How often is the model registry updated?
The registry refreshes every 6 hours from models.dev, OpenRouter, and provider APIs.
Can I use this in CI/CD pipelines?
Yes. The `mm ci` command exits with non-zero status when models are on a retirement clock, making it suitable for pre-commit hooks and CI checks via `--fail-on retiring` flag.