ariobarin/which-llm

which-llm

Use this skill when model knowledge may be stale. It queries a checked-in Artificial Analysis plus OpenRouter snapshot and can refresh it on demand.

Workflow

1. Run commands from this directory with python.
2. If freshness matters, run python query.py data status. If the snapshot is stale, run python query.py data refresh.
3. Use the narrowest command:
   • python query.py models [pattern] [filters] for shortlists.
   • python query.py compare <model>... for side-by-side comparisons.
   • python query.py slug <model> for OpenRouter endpoint names.
   • python query.py show <model> before recommending a specific model.
4. Explain cost fields correctly:
   • idx-run$ is the estimated cost to run the AA benchmark suite.
   • idx-tok is total benchmark-run token use.
   • in$/1m and out$/1m are API prices per million tokens.
5. Prefer openrouter_slug for production. Mention openrouter_free_slug only as a prototype option because :free endpoints can be rate-limited or served differently.

Fast Recipes

python query.py models --intel-min 50 --reasoning --sort cost --top 8
python query.py models --modality text,image --max-cost 500 --sort intel --top 8
python query.py models --no-reasoning --max-latency 6 --sort intel --top 8
python query.py models --context-min 256000 --sort cost --top 8
python query.py models --open-weights --sort intel --top 8
python query.py models --free --sort cost --top 20
python query.py compare claude-opus-4-7 gpt-5 gemini-3-1-pro
python query.py slug claude-opus-4-7

Use python query.py models --help for all filters, including --json.

Do Not Use For

• Domain evals or private benchmarks that AA does not track.
• Models so new that AA has not indexed them yet.
• Authoritative non-OpenRouter provider pricing. Verify those prices with the provider.