Skip to content

vexy-co-model-catalog

vexy-co-model-catalog downloads the list of available models from AI providers, writes each list to disk in a stable, sorted form, and keeps track of which providers answered and which failed. It ships a command-line tool, vmc (also installed as vexy-model-catalog), and a Python package, vexy_co_model_catalog.

What goes in, what comes out

You give the tool a provider name, for example openai, and an API key in the environment variable that provider expects, for example OPENAI_API_KEY.

The tool then:

  1. Looks up the provider in its built-in table of 40 providers. The table holds the provider type, the API key variable and the variable that holds the base URL.
  2. Sends one GET request to <base URL>/models with the right authentication header (Authorization: Bearer … for OpenAI-compatible APIs, x-api-key for Anthropic).
  3. Sorts the response by model ID.
  4. Writes two files into the current directory:
    • config/json/models_openai.json — the full response, sorted.
    • config/txt/models_openai.txt — one model ID per line.
  5. Records the result in models/extra/failed_providers.json, so a later run can skip providers that are failing.
export OPENAI_API_KEY="sk-..."
vmc fetch openai

✓ openai: <N> models

Fetch Complete: 1 successful, 0 failed, 0 skipped

<N> is the number of model IDs the provider returned.

Who it is for

  • People who maintain configuration for several LLM command-line tools and want a current list of model IDs per provider.
  • Scripts that need a local, versionable snapshot of what each provider offers.
  • Users of the older dump_models.py script, which this package replaces. The file names (models_<provider>.json, models_<provider>.txt) are the same.

Where to go next

If you want to… Read
Install the tool Installation
Fetch your first catalog Quick start
Look up a command or flag CLI reference
See which providers exist and which variables they read Providers
Understand the files the tool writes Output files
Control logging, analytics and caching Configuration
Call the library from Python Python API
Move from dump_models.py Migrating from dump_models.py
Fix a problem Troubleshooting
Work on the code Development

Status

The package is versioned from git tags (v1.0.0 to v1.0.4 exist) and is not yet published on PyPI. Until it is, install it from the GitHub repository; see Installation.