A single-file, zero-config CLI for exploring and testing the entire OpenRouter catalog — not just chat models. It discovers every modality OpenRouter exposes (LLM, embeddings, reranker, image, video, TTS, transcription) and fires a real request at whichever model you pick, using the correct API for that modality.
OpenRouter's catalog is huge and heterogeneous. Comparing providers normally means reading docs and writing a different test script per modality. This tool collapses that into one interactive session: browse → filter → test → export.
- Fetch the full OpenRouter model catalog
- Merge dedicated image / video / embedding catalogs
- Auto-discover TTS, transcription, rerank and other modality endpoints
- Search models, filter by capability, inspect raw metadata
- Per-provider endpoint details (which providers serve a model, pricing, context)
- Test any model with the right API for its type
- LLM / chat · embeddings · reranker
- image generation · video generation (with job polling)
- TTS · transcription
- Check credit balance
- Export the catalog to JSON or CSV
- Test outputs (images, audio, transcripts) saved beside the script — never overwritten (
_1,_2, … suffixes)
Python 3.8+, one dependency:
pip install -r requirements.txtexport OPENROUTER_API_KEY="sk-or-v1-..." # Windows: set OPENROUTER_API_KEY=sk-or-v1-...
python openrouter_explorer.pyNo key set? You get a hidden getpass prompt instead — the key is never printed, logged, or stored.
1. Summary 6. Provider endpoint details
2. List models 7. TEST MODEL
3. Search models 8. Check credit balance
4. Filter by capability 9. Export JSON
5. Inspect model JSON 10. Export CSV
0. Exit
- Single file. Stdlib +
requests. No config files, no build step. OPENROUTER_BASE_URLoverrides the API root (defaulthttps://openrouter.ai/api/v1).- Exports and test artifacts are written next to the script.